Debian Codegraph @

rfc1123.html

raw
1 <pre>Network Working Group Internet Engineering Task Force
2 Request for Comments: 1123 R. Braden, Editor
3 October 1989
4
5
6 <span class="h1">Requirements for Internet Hosts -- Application and Support</span>
7
8 Status of This Memo
9
10 This RFC is an official specification for the Internet community. It
11 incorporates by reference, amends, corrects, and supplements the
12 primary protocol standards documents relating to hosts. Distribution
13 of this document is unlimited.
14
15 Summary
16
17 This RFC is one of a pair that defines and discusses the requirements
18 for Internet host software. This RFC covers the application and
19 support protocols; its companion <a href="./rfc1122">RFC-1122</a> covers the communication
20 protocol layers: link layer, IP layer, and transport layer.
21
22
23
24 Table of Contents
25
26
27
28
29 <a href="#section-1">1</a>. INTRODUCTION ............................................... <a href="#page-5">5</a>
30 <a href="#section-1.1">1.1</a> The Internet Architecture .............................. <a href="#page-6">6</a>
31 <a href="#section-1.2">1.2</a> General Considerations ................................. <a href="#page-6">6</a>
32 <a href="#section-1.2.1">1.2.1</a> Continuing Internet Evolution ..................... <a href="#page-6">6</a>
33 <a href="#section-1.2.2">1.2.2</a> Robustness Principle .............................. <a href="#page-7">7</a>
34 <a href="#section-1.2.3">1.2.3</a> Error Logging ..................................... <a href="#page-8">8</a>
35 <a href="#section-1.2.4">1.2.4</a> Configuration ..................................... <a href="#page-8">8</a>
36 <a href="#section-1.3">1.3</a> Reading this Document .................................. <a href="#page-10">10</a>
37 <a href="#section-1.3.1">1.3.1</a> Organization ...................................... <a href="#page-10">10</a>
38 <a href="#section-1.3.2">1.3.2</a> Requirements ...................................... <a href="#page-10">10</a>
39 <a href="#section-1.3.3">1.3.3</a> Terminology ....................................... <a href="#page-11">11</a>
40 <a href="#section-1.4">1.4</a> Acknowledgments ........................................ <a href="#page-12">12</a>
41
42 <a href="#section-2">2</a>. GENERAL ISSUES ............................................. <a href="#page-13">13</a>
43 <a href="#section-2.1">2.1</a> Host Names and Numbers ................................. <a href="#page-13">13</a>
44 <a href="#section-2.2">2.2</a> Using Domain Name Service .............................. <a href="#page-13">13</a>
45 <a href="#section-2.3">2.3</a> Applications on Multihomed hosts ....................... <a href="#page-14">14</a>
46 <a href="#section-2.4">2.4</a> Type-of-Service ........................................ <a href="#page-14">14</a>
47 <a href="#section-2.5">2.5</a> GENERAL APPLICATION REQUIREMENTS SUMMARY ............... <a href="#page-15">15</a>
48
49
50
51
52 <span class="grey">Internet Engineering Task Force [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
55
56
57 <a href="#section-3">3</a>. REMOTE LOGIN -- TELNET PROTOCOL ............................ <a href="#page-16">16</a>
58 <a href="#section-3.1">3.1</a> INTRODUCTION ........................................... <a href="#page-16">16</a>
59 <a href="#section-3.2">3.2</a> PROTOCOL WALK-THROUGH .................................. <a href="#page-16">16</a>
60 <a href="#section-3.2.1">3.2.1</a> Option Negotiation ................................ <a href="#page-16">16</a>
61 <a href="#section-3.2.2">3.2.2</a> Telnet Go-Ahead Function .......................... <a href="#page-16">16</a>
62 <a href="#section-3.2.3">3.2.3</a> Control Functions ................................. <a href="#page-17">17</a>
63 <a href="#section-3.2.4">3.2.4</a> Telnet "Synch" Signal ............................. <a href="#page-18">18</a>
64 <a href="#section-3.2.5">3.2.5</a> NVT Printer and Keyboard .......................... <a href="#page-19">19</a>
65 <a href="#section-3.2.6">3.2.6</a> Telnet Command Structure .......................... <a href="#page-20">20</a>
66 <a href="#section-3.2.7">3.2.7</a> Telnet Binary Option .............................. <a href="#page-20">20</a>
67 <a href="#section-3.2.8">3.2.8</a> Telnet Terminal-Type Option ....................... <a href="#page-20">20</a>
68 <a href="#section-3.3">3.3</a> SPECIFIC ISSUES ........................................ <a href="#page-21">21</a>
69 <a href="#section-3.3.1">3.3.1</a> Telnet End-of-Line Convention ..................... <a href="#page-21">21</a>
70 <a href="#section-3.3.2">3.3.2</a> Data Entry Terminals .............................. <a href="#page-23">23</a>
71 <a href="#section-3.3.3">3.3.3</a> Option Requirements ............................... <a href="#page-24">24</a>
72 <a href="#section-3.3.4">3.3.4</a> Option Initiation ................................. <a href="#page-24">24</a>
73 <a href="#section-3.3.5">3.3.5</a> Telnet Linemode Option ............................ <a href="#page-25">25</a>
74 <a href="#section-3.4">3.4</a> TELNET/USER INTERFACE .................................. <a href="#page-25">25</a>
75 <a href="#section-3.4.1">3.4.1</a> Character Set Transparency ........................ <a href="#page-25">25</a>
76 <a href="#section-3.4.2">3.4.2</a> Telnet Commands ................................... <a href="#page-26">26</a>
77 <a href="#section-3.4.3">3.4.3</a> TCP Connection Errors ............................. <a href="#page-26">26</a>
78 <a href="#section-3.4.4">3.4.4</a> Non-Default Telnet Contact Port ................... <a href="#page-26">26</a>
79 <a href="#section-3.4.5">3.4.5</a> Flushing Output ................................... <a href="#page-26">26</a>
80 <a href="#section-3.5">3.5</a>. TELNET REQUIREMENTS SUMMARY ........................... <a href="#page-27">27</a>
81
82 <a href="#section-4">4</a>. FILE TRANSFER .............................................. <a href="#page-29">29</a>
83 <a href="#section-4.1">4.1</a> FILE TRANSFER PROTOCOL -- FTP .......................... <a href="#page-29">29</a>
84 <a href="#section-4.1.1">4.1.1</a> INTRODUCTION ...................................... <a href="#page-29">29</a>
85 <a href="#section-4.1.2">4.1.2</a>. PROTOCOL WALK-THROUGH ............................ <a href="#page-29">29</a>
86 <a href="#section-4.1.2.1">4.1.2.1</a> LOCAL Type ................................... <a href="#page-29">29</a>
87 <a href="#section-4.1.2.2">4.1.2.2</a> Telnet Format Control ........................ <a href="#page-30">30</a>
88 <a href="#section-4.1.2.3">4.1.2.3</a> Page Structure ............................... <a href="#page-30">30</a>
89 <a href="#section-4.1.2.4">4.1.2.4</a> Data Structure Transformations ............... <a href="#page-30">30</a>
90 <a href="#section-4.1.2.5">4.1.2.5</a> Data Connection Management ................... <a href="#page-31">31</a>
91 <a href="#section-4.1.2.6">4.1.2.6</a> PASV Command ................................. <a href="#page-31">31</a>
92 <a href="#section-4.1.2.7">4.1.2.7</a> LIST and NLST Commands ....................... <a href="#page-31">31</a>
93 <a href="#section-4.1.2.8">4.1.2.8</a> SITE Command ................................. <a href="#page-32">32</a>
94 <a href="#section-4.1.2.9">4.1.2.9</a> STOU Command ................................. <a href="#page-32">32</a>
95 <a href="#section-4.1.2.10">4.1.2.10</a> Telnet End-of-line Code ..................... <a href="#page-32">32</a>
96 <a href="#section-4.1.2.11">4.1.2.11</a> FTP Replies ................................. <a href="#page-33">33</a>
97 <a href="#section-4.1.2.12">4.1.2.12</a> Connections ................................. <a href="#page-34">34</a>
98 <a href="#section-4.1.2.13">4.1.2.13</a> Minimum Implementation; <a href="./rfc959">RFC-959</a> Section ..... <a href="#page-34">34</a>
99 <a href="#section-4.1.3">4.1.3</a> SPECIFIC ISSUES ................................... <a href="#page-35">35</a>
100 <a href="#section-4.1.3.1">4.1.3.1</a> Non-standard Command Verbs ................... <a href="#page-35">35</a>
101 <a href="#section-4.1.3.2">4.1.3.2</a> Idle Timeout ................................. <a href="#page-36">36</a>
102 <a href="#section-4.1.3.3">4.1.3.3</a> Concurrency of Data and Control .............. <a href="#page-36">36</a>
103 <a href="#section-4.1.3.4">4.1.3.4</a> FTP Restart Mechanism ........................ <a href="#page-36">36</a>
104 <a href="#section-4.1.4">4.1.4</a> FTP/USER INTERFACE ................................ <a href="#page-39">39</a>
105
106
107
108 <span class="grey">Internet Engineering Task Force [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
111
112
113 <a href="#section-4.1.4.1">4.1.4.1</a> Pathname Specification ....................... <a href="#page-39">39</a>
114 <a href="#section-4.1.4.2">4.1.4.2</a> "QUOTE" Command .............................. <a href="#page-40">40</a>
115 <a href="#section-4.1.4.3">4.1.4.3</a> Displaying Replies to User ................... <a href="#page-40">40</a>
116 <a href="#section-4.1.4.4">4.1.4.4</a> Maintaining Synchronization .................. <a href="#page-40">40</a>
117 <a href="#section-4.1.5">4.1.5</a> FTP REQUIREMENTS SUMMARY ......................... <a href="#page-41">41</a>
118 <a href="#section-4.2">4.2</a> TRIVIAL FILE TRANSFER PROTOCOL -- TFTP ................. <a href="#page-44">44</a>
119 <a href="#section-4.2.1">4.2.1</a> INTRODUCTION ...................................... <a href="#page-44">44</a>
120 <a href="#section-4.2.2">4.2.2</a> PROTOCOL WALK-THROUGH ............................. <a href="#page-44">44</a>
121 <a href="#section-4.2.2.1">4.2.2.1</a> Transfer Modes ............................... <a href="#page-44">44</a>
122 <a href="#section-4.2.2.2">4.2.2.2</a> UDP Header ................................... <a href="#page-44">44</a>
123 <a href="#section-4.2.3">4.2.3</a> SPECIFIC ISSUES ................................... <a href="#page-44">44</a>
124 <a href="#section-4.2.3.1">4.2.3.1</a> Sorcerer's Apprentice Syndrome ............... <a href="#page-44">44</a>
125 <a href="#section-4.2.3.2">4.2.3.2</a> Timeout Algorithms ........................... <a href="#page-46">46</a>
126 <a href="#section-4.2.3.3">4.2.3.3</a> Extensions ................................... <a href="#page-46">46</a>
127 <a href="#section-4.2.3.4">4.2.3.4</a> Access Control ............................... <a href="#page-46">46</a>
128 <a href="#section-4.2.3.5">4.2.3.5</a> Broadcast Request ............................ <a href="#page-46">46</a>
129 <a href="#section-4.2.4">4.2.4</a> TFTP REQUIREMENTS SUMMARY ......................... <a href="#page-47">47</a>
130
131 <a href="#section-5">5</a>. ELECTRONIC MAIL -- SMTP and <a href="./rfc822">RFC-822</a> ........................ <a href="#page-48">48</a>
132 <a href="#section-5.1">5.1</a> INTRODUCTION ........................................... <a href="#page-48">48</a>
133 <a href="#section-5.2">5.2</a> PROTOCOL WALK-THROUGH .................................. <a href="#page-48">48</a>
134 <a href="#section-5.2.1">5.2.1</a> The SMTP Model .................................... <a href="#page-48">48</a>
135 <a href="#section-5.2.2">5.2.2</a> Canonicalization .................................. <a href="#page-49">49</a>
136 <a href="#section-5.2.3">5.2.3</a> VRFY and EXPN Commands ............................ <a href="#page-50">50</a>
137 <a href="#section-5.2.4">5.2.4</a> SEND, SOML, and SAML Commands ..................... <a href="#page-50">50</a>
138 <a href="#section-5.2.5">5.2.5</a> HELO Command ...................................... <a href="#page-50">50</a>
139 <a href="#section-5.2.6">5.2.6</a> Mail Relay ........................................ <a href="#page-51">51</a>
140 <a href="#section-5.2.7">5.2.7</a> RCPT Command ...................................... <a href="#page-52">52</a>
141 <a href="#section-5.2.8">5.2.8</a> DATA Command ...................................... <a href="#page-53">53</a>
142 <a href="#section-5.2.9">5.2.9</a> Command Syntax .................................... <a href="#page-54">54</a>
143 <a href="#section-5.2.10">5.2.10</a> SMTP Replies ..................................... <a href="#page-54">54</a>
144 <a href="#section-5.2.11">5.2.11</a> Transparency ..................................... <a href="#page-55">55</a>
145 <a href="#section-5.2.12">5.2.12</a> WKS Use in MX Processing ......................... <a href="#page-55">55</a>
146 <a href="#section-5.2.13">5.2.13</a> <a href="./rfc822">RFC-822</a> Message Specification .................... <a href="#page-55">55</a>
147 <a href="#section-5.2.14">5.2.14</a> <a href="./rfc822">RFC-822</a> Date and Time Specification .............. <a href="#page-55">55</a>
148 <a href="#section-5.2.15">5.2.15</a> <a href="./rfc822">RFC-822</a> Syntax Change ............................ <a href="#page-56">56</a>
149 <a href="#section-5.2.16">5.2.16</a> <a href="./rfc822">RFC-822</a> Local-part .............................. <a href="#page-56">56</a>
150 <a href="#section-5.2.17">5.2.17</a> Domain Literals .................................. <a href="#page-57">57</a>
151 <a href="#section-5.2.18">5.2.18</a> Common Address Formatting Errors ................. <a href="#page-58">58</a>
152 <a href="#section-5.2.19">5.2.19</a> Explicit Source Routes ........................... <a href="#page-58">58</a>
153 <a href="#section-5.3">5.3</a> SPECIFIC ISSUES ........................................ <a href="#page-59">59</a>
154 <a href="#section-5.3.1">5.3.1</a> SMTP Queueing Strategies .......................... <a href="#page-59">59</a>
155 <a href="#section-5.3.1.1">5.3.1.1</a> Sending Strategy .............................. <a href="#page-59">59</a>
156 <a href="#section-5.3.1.2">5.3.1.2</a> Receiving strategy ........................... <a href="#page-61">61</a>
157 <a href="#section-5.3.2">5.3.2</a> Timeouts in SMTP .................................. <a href="#page-61">61</a>
158 <a href="#section-5.3.3">5.3.3</a> Reliable Mail Receipt ............................. <a href="#page-63">63</a>
159 <a href="#section-5.3.4">5.3.4</a> Reliable Mail Transmission ........................ <a href="#page-63">63</a>
160 <a href="#section-5.3.5">5.3.5</a> Domain Name Support ............................... <a href="#page-65">65</a>
161
162
163
164 <span class="grey">Internet Engineering Task Force [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
167
168
169 <a href="#section-5.3.6">5.3.6</a> Mailing Lists and Aliases ......................... <a href="#page-65">65</a>
170 <a href="#section-5.3.7">5.3.7</a> Mail Gatewaying ................................... <a href="#page-66">66</a>
171 <a href="#section-5.3.8">5.3.8</a> Maximum Message Size .............................. <a href="#page-68">68</a>
172 <a href="#section-5.4">5.4</a> SMTP REQUIREMENTS SUMMARY .............................. <a href="#page-69">69</a>
173
174 <a href="#section-6">6</a>. SUPPORT SERVICES ............................................ <a href="#page-72">72</a>
175 <a href="#section-6.1">6.1</a> DOMAIN NAME TRANSLATION ................................. <a href="#page-72">72</a>
176 <a href="#section-6.1.1">6.1.1</a> INTRODUCTION ....................................... <a href="#page-72">72</a>
177 <a href="#section-6.1.2">6.1.2</a> PROTOCOL WALK-THROUGH ............................. <a href="#page-72">72</a>
178 <a href="#section-6.1.2.1">6.1.2.1</a> Resource Records with Zero TTL ............... <a href="#page-73">73</a>
179 <a href="#section-6.1.2.2">6.1.2.2</a> QCLASS Values ................................ <a href="#page-73">73</a>
180 <a href="#section-6.1.2.3">6.1.2.3</a> Unused Fields ................................ <a href="#page-73">73</a>
181 <a href="#section-6.1.2.4">6.1.2.4</a> Compression .................................. <a href="#page-73">73</a>
182 <a href="#section-6.1.2.5">6.1.2.5</a> Misusing Configuration Info .................. <a href="#page-73">73</a>
183 <a href="#section-6.1.3">6.1.3</a> SPECIFIC ISSUES ................................... <a href="#page-74">74</a>
184 <a href="#section-6.1.3.1">6.1.3.1</a> Resolver Implementation ...................... <a href="#page-74">74</a>
185 <a href="#section-6.1.3.2">6.1.3.2</a> Transport Protocols .......................... <a href="#page-75">75</a>
186 <a href="#section-6.1.3.3">6.1.3.3</a> Efficient Resource Usage ..................... <a href="#page-77">77</a>
187 <a href="#section-6.1.3.4">6.1.3.4</a> Multihomed Hosts ............................. <a href="#page-78">78</a>
188 <a href="#section-6.1.3.5">6.1.3.5</a> Extensibility ................................ <a href="#page-79">79</a>
189 <a href="#section-6.1.3.6">6.1.3.6</a> Status of RR Types ........................... <a href="#page-79">79</a>
190 <a href="#section-6.1.3.7">6.1.3.7</a> Robustness ................................... <a href="#page-80">80</a>
191 <a href="#section-6.1.3.8">6.1.3.8</a> Local Host Table ............................. <a href="#page-80">80</a>
192 <a href="#section-6.1.4">6.1.4</a> DNS USER INTERFACE ................................ <a href="#page-81">81</a>
193 <a href="#section-6.1.4.1">6.1.4.1</a> DNS Administration ........................... <a href="#page-81">81</a>
194 <a href="#section-6.1.4.2">6.1.4.2</a> DNS User Interface ........................... <a href="#page-81">81</a>
195 <a href="#section-6.1.4.3">6.1.4.3</a> Interface Abbreviation Facilities ............. <a href="#page-82">82</a>
196 <a href="#section-6.1.5">6.1.5</a> DOMAIN NAME SYSTEM REQUIREMENTS SUMMARY ........... <a href="#page-84">84</a>
197 <a href="#section-6.2">6.2</a> HOST INITIALIZATION .................................... <a href="#page-87">87</a>
198 <a href="#section-6.2.1">6.2.1</a> INTRODUCTION ...................................... <a href="#page-87">87</a>
199 <a href="#section-6.2.2">6.2.2</a> REQUIREMENTS ...................................... <a href="#page-87">87</a>
200 <a href="#section-6.2.2.1">6.2.2.1</a> Dynamic Configuration ........................ <a href="#page-87">87</a>
201 <a href="#section-6.2.2.2">6.2.2.2</a> Loading Phase ................................ <a href="#page-89">89</a>
202 <a href="#section-6.3">6.3</a> REMOTE MANAGEMENT ...................................... <a href="#page-90">90</a>
203 <a href="#section-6.3.1">6.3.1</a> INTRODUCTION ...................................... <a href="#page-90">90</a>
204 <a href="#section-6.3.2">6.3.2</a> PROTOCOL WALK-THROUGH ............................. <a href="#page-90">90</a>
205 <a href="#section-6.3.3">6.3.3</a> MANAGEMENT REQUIREMENTS SUMMARY ................... <a href="#page-92">92</a>
206
207 <a href="#section-7">7</a>. REFERENCES ................................................. <a href="#page-93">93</a>
208
209
210
211
212
213
214
215
216
217
218
219
220 <span class="grey">Internet Engineering Task Force [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
223
224
225 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. INTRODUCTION</span>
226
227 This document is one of a pair that defines and discusses the
228 requirements for host system implementations of the Internet protocol
229 suite. This RFC covers the applications layer and support protocols.
230 Its companion RFC, "Requirements for Internet Hosts -- Communications
231 Layers" [INTRO:1] covers the lower layer protocols: transport layer,
232 IP layer, and link layer.
233
234 These documents are intended to provide guidance for vendors,
235 implementors, and users of Internet communication software. They
236 represent the consensus of a large body of technical experience and
237 wisdom, contributed by members of the Internet research and vendor
238 communities.
239
240 This RFC enumerates standard protocols that a host connected to the
241 Internet must use, and it incorporates by reference the RFCs and
242 other documents describing the current specifications for these
243 protocols. It corrects errors in the referenced documents and adds
244 additional discussion and guidance for an implementor.
245
246 For each protocol, this document also contains an explicit set of
247 requirements, recommendations, and options. The reader must
248 understand that the list of requirements in this document is
249 incomplete by itself; the complete set of requirements for an
250 Internet host is primarily defined in the standard protocol
251 specification documents, with the corrections, amendments, and
252 supplements contained in this RFC.
253
254 A good-faith implementation of the protocols that was produced after
255 careful reading of the RFC's and with some interaction with the
256 Internet technical community, and that followed good communications
257 software engineering practices, should differ from the requirements
258 of this document in only minor ways. Thus, in many cases, the
259 "requirements" in this RFC are already stated or implied in the
260 standard protocol documents, so that their inclusion here is, in a
261 sense, redundant. However, they were included because some past
262 implementation has made the wrong choice, causing problems of
263 interoperability, performance, and/or robustness.
264
265 This document includes discussion and explanation of many of the
266 requirements and recommendations. A simple list of requirements
267 would be dangerous, because:
268
269 o Some required features are more important than others, and some
270 features are optional.
271
272 o There may be valid reasons why particular vendor products that
273
274
275
276 <span class="grey">Internet Engineering Task Force [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
279
280
281 are designed for restricted contexts might choose to use
282 different specifications.
283
284 However, the specifications of this document must be followed to meet
285 the general goal of arbitrary host interoperation across the
286 diversity and complexity of the Internet system. Although most
287 current implementations fail to meet these requirements in various
288 ways, some minor and some major, this specification is the ideal
289 towards which we need to move.
290
291 These requirements are based on the current level of Internet
292 architecture. This document will be updated as required to provide
293 additional clarifications or to include additional information in
294 those areas in which specifications are still evolving.
295
296 This introductory section begins with general advice to host software
297 vendors, and then gives some guidance on reading the rest of the
298 document. <a href="#section-2">Section 2</a> contains general requirements that may be
299 applicable to all application and support protocols. Sections <a href="#section-3">3</a>, <a href="#section-4">4</a>,
300 and 5 contain the requirements on protocols for the three major
301 applications: Telnet, file transfer, and electronic mail,
302 respectively. <a href="#section-6">Section 6</a> covers the support applications: the domain
303 name system, system initialization, and management. Finally, all
304 references will be found in <a href="#section-7">Section 7</a>.
305
306 1.1 The Internet Architecture
307
308 For a brief introduction to the Internet architecture from a host
309 viewpoint, see <a href="#section-1.1">Section 1.1</a> of [INTRO:1]. That section also
310 contains recommended references for general background on the
311 Internet architecture.
312
313 1.2 General Considerations
314
315 There are two important lessons that vendors of Internet host
316 software have learned and which a new vendor should consider
317 seriously.
318
319 1.2.1 Continuing Internet Evolution
320
321 The enormous growth of the Internet has revealed problems of
322 management and scaling in a large datagram-based packet
323 communication system. These problems are being addressed, and
324 as a result there will be continuing evolution of the
325 specifications described in this document. These changes will
326 be carefully planned and controlled, since there is extensive
327 participation in this planning by the vendors and by the
328 organizations responsible for operations of the networks.
329
330
331
332 <span class="grey">Internet Engineering Task Force [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
335
336
337 Development, evolution, and revision are characteristic of
338 computer network protocols today, and this situation will
339 persist for some years. A vendor who develops computer
340 communication software for the Internet protocol suite (or any
341 other protocol suite!) and then fails to maintain and update
342 that software for changing specifications is going to leave a
343 trail of unhappy customers. The Internet is a large
344 communication network, and the users are in constant contact
345 through it. Experience has shown that knowledge of
346 deficiencies in vendor software propagates quickly through the
347 Internet technical community.
348
349 1.2.2 Robustness Principle
350
351 At every layer of the protocols, there is a general rule whose
352 application can lead to enormous benefits in robustness and
353 interoperability:
354
355 "Be liberal in what you accept, and
356 conservative in what you send"
357
358 Software should be written to deal with every conceivable
359 error, no matter how unlikely; sooner or later a packet will
360 come in with that particular combination of errors and
361 attributes, and unless the software is prepared, chaos can
362 ensue. In general, it is best to assume that the network is
363 filled with malevolent entities that will send in packets
364 designed to have the worst possible effect. This assumption
365 will lead to suitable protective design, although the most
366 serious problems in the Internet have been caused by
367 unenvisaged mechanisms triggered by low-probability events;
368 mere human malice would never have taken so devious a course!
369
370 Adaptability to change must be designed into all levels of
371 Internet host software. As a simple example, consider a
372 protocol specification that contains an enumeration of values
373 for a particular header field -- e.g., a type field, a port
374 number, or an error code; this enumeration must be assumed to
375 be incomplete. Thus, if a protocol specification defines four
376 possible error codes, the software must not break when a fifth
377 code shows up. An undefined code might be logged (see below),
378 but it must not cause a failure.
379
380 The second part of the principle is almost as important:
381 software on other hosts may contain deficiencies that make it
382 unwise to exploit legal but obscure protocol features. It is
383 unwise to stray far from the obvious and simple, lest untoward
384 effects result elsewhere. A corollary of this is "watch out
385
386
387
388 <span class="grey">Internet Engineering Task Force [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
391
392
393 for misbehaving hosts"; host software should be prepared, not
394 just to survive other misbehaving hosts, but also to cooperate
395 to limit the amount of disruption such hosts can cause to the
396 shared communication facility.
397
398 1.2.3 Error Logging
399
400 The Internet includes a great variety of host and gateway
401 systems, each implementing many protocols and protocol layers,
402 and some of these contain bugs and mis-features in their
403 Internet protocol software. As a result of complexity,
404 diversity, and distribution of function, the diagnosis of user
405 problems is often very difficult.
406
407 Problem diagnosis will be aided if host implementations include
408 a carefully designed facility for logging erroneous or
409 "strange" protocol events. It is important to include as much
410 diagnostic information as possible when an error is logged. In
411 particular, it is often useful to record the header(s) of a
412 packet that caused an error. However, care must be taken to
413 ensure that error logging does not consume prohibitive amounts
414 of resources or otherwise interfere with the operation of the
415 host.
416
417 There is a tendency for abnormal but harmless protocol events
418 to overflow error logging files; this can be avoided by using a
419 "circular" log, or by enabling logging only while diagnosing a
420 known failure. It may be useful to filter and count duplicate
421 successive messages. One strategy that seems to work well is:
422 (1) always count abnormalities and make such counts accessible
423 through the management protocol (see <a href="#section-6.3">Section 6.3</a>); and (2)
424 allow the logging of a great variety of events to be
425 selectively enabled. For example, it might useful to be able
426 to "log everything" or to "log everything for host X".
427
428 Note that different managements may have differing policies
429 about the amount of error logging that they want normally
430 enabled in a host. Some will say, "if it doesn't hurt me, I
431 don't want to know about it", while others will want to take a
432 more watchful and aggressive attitude about detecting and
433 removing protocol abnormalities.
434
435 1.2.4 Configuration
436
437 It would be ideal if a host implementation of the Internet
438 protocol suite could be entirely self-configuring. This would
439 allow the whole suite to be implemented in ROM or cast into
440 silicon, it would simplify diskless workstations, and it would
441
442
443
444 <span class="grey">Internet Engineering Task Force [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
447
448
449 be an immense boon to harried LAN administrators as well as
450 system vendors. We have not reached this ideal; in fact, we
451 are not even close.
452
453 At many points in this document, you will find a requirement
454 that a parameter be a configurable option. There are several
455 different reasons behind such requirements. In a few cases,
456 there is current uncertainty or disagreement about the best
457 value, and it may be necessary to update the recommended value
458 in the future. In other cases, the value really depends on
459 external factors -- e.g., the size of the host and the
460 distribution of its communication load, or the speeds and
461 topology of nearby networks -- and self-tuning algorithms are
462 unavailable and may be insufficient. In some cases,
463 configurability is needed because of administrative
464 requirements.
465
466 Finally, some configuration options are required to communicate
467 with obsolete or incorrect implementations of the protocols,
468 distributed without sources, that unfortunately persist in many
469 parts of the Internet. To make correct systems coexist with
470 these faulty systems, administrators often have to "mis-
471 configure" the correct systems. This problem will correct
472 itself gradually as the faulty systems are retired, but it
473 cannot be ignored by vendors.
474
475 When we say that a parameter must be configurable, we do not
476 intend to require that its value be explicitly read from a
477 configuration file at every boot time. We recommend that
478 implementors set up a default for each parameter, so a
479 configuration file is only necessary to override those defaults
480 that are inappropriate in a particular installation. Thus, the
481 configurability requirement is an assurance that it will be
482 POSSIBLE to override the default when necessary, even in a
483 binary-only or ROM-based product.
484
485 This document requires a particular value for such defaults in
486 some cases. The choice of default is a sensitive issue when
487 the configuration item controls the accommodation to existing
488 faulty systems. If the Internet is to converge successfully to
489 complete interoperability, the default values built into
490 implementations must implement the official protocol, not
491 "mis-configurations" to accommodate faulty implementations.
492 Although marketing considerations have led some vendors to
493 choose mis-configuration defaults, we urge vendors to choose
494 defaults that will conform to the standard.
495
496 Finally, we note that a vendor needs to provide adequate
497
498
499
500 <span class="grey">Internet Engineering Task Force [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
503
504
505 documentation on all configuration parameters, their limits and
506 effects.
507
508
509 1.3 Reading this Document
510
511 1.3.1 Organization
512
513 In general, each major section is organized into the following
514 subsections:
515
516 (1) Introduction
517
518 (2) Protocol Walk-Through -- considers the protocol
519 specification documents section-by-section, correcting
520 errors, stating requirements that may be ambiguous or
521 ill-defined, and providing further clarification or
522 explanation.
523
524 (3) Specific Issues -- discusses protocol design and
525 implementation issues that were not included in the walk-
526 through.
527
528 (4) Interfaces -- discusses the service interface to the next
529 higher layer.
530
531 (5) Summary -- contains a summary of the requirements of the
532 section.
533
534 Under many of the individual topics in this document, there is
535 parenthetical material labeled "DISCUSSION" or
536 "IMPLEMENTATION". This material is intended to give
537 clarification and explanation of the preceding requirements
538 text. It also includes some suggestions on possible future
539 directions or developments. The implementation material
540 contains suggested approaches that an implementor may want to
541 consider.
542
543 The summary sections are intended to be guides and indexes to
544 the text, but are necessarily cryptic and incomplete. The
545 summaries should never be used or referenced separately from
546 the complete RFC.
547
548 1.3.2 Requirements
549
550 In this document, the words that are used to define the
551 significance of each particular requirement are capitalized.
552 These words are:
553
554
555
556 <span class="grey">Internet Engineering Task Force [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
559
560
561 * "MUST"
562
563 This word or the adjective "REQUIRED" means that the item
564 is an absolute requirement of the specification.
565
566 * "SHOULD"
567
568 This word or the adjective "RECOMMENDED" means that there
569 may exist valid reasons in particular circumstances to
570 ignore this item, but the full implications should be
571 understood and the case carefully weighed before choosing
572 a different course.
573
574 * "MAY"
575
576 This word or the adjective "OPTIONAL" means that this item
577 is truly optional. One vendor may choose to include the
578 item because a particular marketplace requires it or
579 because it enhances the product, for example; another
580 vendor may omit the same item.
581
582
583 An implementation is not compliant if it fails to satisfy one
584 or more of the MUST requirements for the protocols it
585 implements. An implementation that satisfies all the MUST and
586 all the SHOULD requirements for its protocols is said to be
587 "unconditionally compliant"; one that satisfies all the MUST
588 requirements but not all the SHOULD requirements for its
589 protocols is said to be "conditionally compliant".
590
591 1.3.3 Terminology
592
593 This document uses the following technical terms:
594
595 Segment
596 A segment is the unit of end-to-end transmission in the
597 TCP protocol. A segment consists of a TCP header followed
598 by application data. A segment is transmitted by
599 encapsulation in an IP datagram.
600
601 Message
602 This term is used by some application layer protocols
603 (particularly SMTP) for an application data unit.
604
605 Datagram
606 A [UDP] datagram is the unit of end-to-end transmission in
607 the UDP protocol.
608
609
610
611
612 <span class="grey">Internet Engineering Task Force [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc1123">RFC1123</a> INTRODUCTION October 1989</span>
615
616
617 Multihomed
618 A host is said to be multihomed if it has multiple IP
619 addresses to connected networks.
620
621
622
623 1.4 Acknowledgments
624
625 This document incorporates contributions and comments from a large
626 group of Internet protocol experts, including representatives of
627 university and research labs, vendors, and government agencies.
628 It was assembled primarily by the Host Requirements Working Group
629 of the Internet Engineering Task Force (IETF).
630
631 The Editor would especially like to acknowledge the tireless
632 dedication of the following people, who attended many long
633 meetings and generated 3 million bytes of electronic mail over the
634 past 18 months in pursuit of this document: Philip Almquist, Dave
635 Borman (Cray Research), Noel Chiappa, Dave Crocker (DEC), Steve
636 Deering (Stanford), Mike Karels (Berkeley), Phil Karn (Bellcore),
637 John Lekashman (NASA), Charles Lynn (BBN), Keith McCloghrie (TWG),
638 Paul Mockapetris (ISI), Thomas Narten (Purdue), Craig Partridge
639 (BBN), Drew Perkins (CMU), and James Van Bokkelen (FTP Software).
640
641 In addition, the following people made major contributions to the
642 effort: Bill Barns (Mitre), Steve Bellovin (AT&amp;T), Mike Brescia
643 (BBN), Ed Cain (DCA), Annette DeSchon (ISI), Martin Gross (DCA),
644 Phill Gross (NRI), Charles Hedrick (Rutgers), Van Jacobson (LBL),
645 John Klensin (MIT), Mark Lottor (SRI), Milo Medin (NASA), Bill
646 Melohn (Sun Microsystems), Greg Minshall (Kinetics), Jeff Mogul
647 (DEC), John Mullen (CMC), Jon Postel (ISI), John Romkey (Epilogue
648 Technology), and Mike StJohns (DCA). The following also made
649 significant contributions to particular areas: Eric Allman
650 (Berkeley), Rob Austein (MIT), Art Berggreen (ACC), Keith Bostic
651 (Berkeley), Vint Cerf (NRI), Wayne Hathaway (NASA), Matt Korn
652 (IBM), Erik Naggum (Naggum Software, Norway), Robert Ullmann
653 (Prime Computer), David Waitzman (BBN), Frank Wancho (USA), Arun
654 Welch (Ohio State), Bill Westfield (Cisco), and Rayan Zachariassen
655 (Toronto).
656
657 We are grateful to all, including any contributors who may have
658 been inadvertently omitted from this list.
659
660
661
662
663
664
665
666
667
668 <span class="grey">Internet Engineering Task Force [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc1123">RFC1123</a> APPLICATIONS LAYER -- GENERAL October 1989</span>
671
672
673 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. GENERAL ISSUES</span>
674
675 This section contains general requirements that may be applicable to
676 all application-layer protocols.
677
678 2.1 Host Names and Numbers
679
680 The syntax of a legal Internet host name was specified in <a href="./rfc952">RFC-952</a>
681 [DNS:4]. One aspect of host name syntax is hereby changed: the
682 restriction on the first character is relaxed to allow either a
683 letter or a digit. Host software MUST support this more liberal
684 syntax.
685
686 Host software MUST handle host names of up to 63 characters and
687 SHOULD handle host names of up to 255 characters.
688
689 Whenever a user inputs the identity of an Internet host, it SHOULD
690 be possible to enter either (1) a host domain name or (2) an IP
691 address in dotted-decimal ("#.#.#.#") form. The host SHOULD check
692 the string syntactically for a dotted-decimal number before
693 looking it up in the Domain Name System.
694
695 DISCUSSION:
696 This last requirement is not intended to specify the complete
697 syntactic form for entering a dotted-decimal host number;
698 that is considered to be a user-interface issue. For
699 example, a dotted-decimal number must be enclosed within
700 "[ ]" brackets for SMTP mail (see <a href="#section-5.2.17">Section 5.2.17</a>). This
701 notation could be made universal within a host system,
702 simplifying the syntactic checking for a dotted-decimal
703 number.
704
705 If a dotted-decimal number can be entered without such
706 identifying delimiters, then a full syntactic check must be
707 made, because a segment of a host domain name is now allowed
708 to begin with a digit and could legally be entirely numeric
709 (see <a href="#section-6.1.2.4">Section 6.1.2.4</a>). However, a valid host name can never
710 have the dotted-decimal form #.#.#.#, since at least the
711 highest-level component label will be alphabetic.
712
713 2.2 Using Domain Name Service
714
715 Host domain names MUST be translated to IP addresses as described
716 in <a href="#section-6.1">Section 6.1</a>.
717
718 Applications using domain name services MUST be able to cope with
719 soft error conditions. Applications MUST wait a reasonable
720 interval between successive retries due to a soft error, and MUST
721
722
723
724 <span class="grey">Internet Engineering Task Force [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc1123">RFC1123</a> APPLICATIONS LAYER -- GENERAL October 1989</span>
727
728
729 allow for the possibility that network problems may deny service
730 for hours or even days.
731
732 An application SHOULD NOT rely on the ability to locate a WKS
733 record containing an accurate listing of all services at a
734 particular host address, since the WKS RR type is not often used
735 by Internet sites. To confirm that a service is present, simply
736 attempt to use it.
737
738 2.3 Applications on Multihomed hosts
739
740 When the remote host is multihomed, the name-to-address
741 translation will return a list of alternative IP addresses. As
742 specified in <a href="#section-6.1.3.4">Section 6.1.3.4</a>, this list should be in order of
743 decreasing preference. Application protocol implementations
744 SHOULD be prepared to try multiple addresses from the list until
745 success is obtained. More specific requirements for SMTP are
746 given in <a href="#section-5.3.4">Section 5.3.4</a>.
747
748 When the local host is multihomed, a UDP-based request/response
749 application SHOULD send the response with an IP source address
750 that is the same as the specific destination address of the UDP
751 request datagram. The "specific destination address" is defined
752 in the "IP Addressing" section of the companion RFC [INTRO:1].
753
754 Similarly, a server application that opens multiple TCP
755 connections to the same client SHOULD use the same local IP
756 address for all.
757
758 2.4 Type-of-Service
759
760 Applications MUST select appropriate TOS values when they invoke
761 transport layer services, and these values MUST be configurable.
762 Note that a TOS value contains 5 bits, of which only the most-
763 significant 3 bits are currently defined; the other two bits MUST
764 be zero.
765
766 DISCUSSION:
767 As gateway algorithms are developed to implement Type-of-
768 Service, the recommended values for various application
769 protocols may change. In addition, it is likely that
770 particular combinations of users and Internet paths will want
771 non-standard TOS values. For these reasons, the TOS values
772 must be configurable.
773
774 See the latest version of the "Assigned Numbers" RFC
775 [INTRO:5] for the recommended TOS values for the major
776 application protocols.
777
778
779
780 <span class="grey">Internet Engineering Task Force [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc1123">RFC1123</a> APPLICATIONS LAYER -- GENERAL October 1989</span>
783
784
785 2.5 GENERAL APPLICATION REQUIREMENTS SUMMARY
786
787 | | | | |S| |
788 | | | | |H| |F
789 | | | | |O|M|o
790 | | |S| |U|U|o
791 | | |H| |L|S|t
792 | |M|O| |D|T|n
793 | |U|U|M| | |o
794 | |S|L|A|N|N|t
795 | |T|D|Y|O|O|t
796 FEATURE |SECTION | | | |T|T|e
797 -----------------------------------------------|----------|-|-|-|-|-|--
798 | | | | | | |
799 User interfaces: | | | | | | |
800 Allow host name to begin with digit |2.1 |x| | | | |
801 Host names of up to 635 characters |2.1 |x| | | | |
802 Host names of up to 255 characters |2.1 | |x| | | |
803 Support dotted-decimal host numbers |2.1 | |x| | | |
804 Check syntactically for dotted-dec first |2.1 | |x| | | |
805 | | | | | | |
806 Map domain names per <a href="#section-6.1">Section 6.1</a> |2.2 |x| | | | |
807 Cope with soft DNS errors |2.2 |x| | | | |
808 Reasonable interval between retries |2.2 |x| | | | |
809 Allow for long outages |2.2 |x| | | | |
810 Expect WKS records to be available |2.2 | | | |x| |
811 | | | | | | |
812 Try multiple addr's for remote multihomed host |2.3 | |x| | | |
813 UDP reply src addr is specific dest of request |2.3 | |x| | | |
814 Use same IP addr for related TCP connections |2.3 | |x| | | |
815 Specify appropriate TOS values |2.4 |x| | | | |
816 TOS values configurable |2.4 |x| | | | |
817 Unused TOS bits zero |2.4 |x| | | | |
818 | | | | | | |
819 | | | | | | |
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836 <span class="grey">Internet Engineering Task Force [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
839
840
841 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. REMOTE LOGIN -- TELNET PROTOCOL</span>
842
843 3.1 INTRODUCTION
844
845 Telnet is the standard Internet application protocol for remote
846 login. It provides the encoding rules to link a user's
847 keyboard/display on a client ("user") system with a command
848 interpreter on a remote server system. A subset of the Telnet
849 protocol is also incorporated within other application protocols,
850 e.g., FTP and SMTP.
851
852 Telnet uses a single TCP connection, and its normal data stream
853 ("Network Virtual Terminal" or "NVT" mode) is 7-bit ASCII with
854 escape sequences to embed control functions. Telnet also allows
855 the negotiation of many optional modes and functions.
856
857 The primary Telnet specification is to be found in <a href="./rfc854">RFC-854</a>
858 [TELNET:1], while the options are defined in many other RFCs; see
859 <a href="#section-7">Section 7</a> for references.
860
861 3.2 PROTOCOL WALK-THROUGH
862
863 3.2.1 Option Negotiation: <a href="./rfc854">RFC-854</a>, pp. 2-3
864
865 Every Telnet implementation MUST include option negotiation and
866 subnegotiation machinery [TELNET:2].
867
868 A host MUST carefully follow the rules of <a href="./rfc854">RFC-854</a> to avoid
869 option-negotiation loops. A host MUST refuse (i.e, reply
870 WONT/DONT to a DO/WILL) an unsupported option. Option
871 negotiation SHOULD continue to function (even if all requests
872 are refused) throughout the lifetime of a Telnet connection.
873
874 If all option negotiations fail, a Telnet implementation MUST
875 default to, and support, an NVT.
876
877 DISCUSSION:
878 Even though more sophisticated "terminals" and supporting
879 option negotiations are becoming the norm, all
880 implementations must be prepared to support an NVT for any
881 user-server communication.
882
883 3.2.2 Telnet Go-Ahead Function: <a href="./rfc854">RFC-854</a>, p. 5, and <a href="./rfc858">RFC-858</a>
884
885 On a host that never sends the Telnet command Go Ahead (GA),
886 the Telnet Server MUST attempt to negotiate the Suppress Go
887 Ahead option (i.e., send "WILL Suppress Go Ahead"). A User or
888 Server Telnet MUST always accept negotiation of the Suppress Go
889
890
891
892 <span class="grey">Internet Engineering Task Force [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
895
896
897 Ahead option.
898
899 When it is driving a full-duplex terminal for which GA has no
900 meaning, a User Telnet implementation MAY ignore GA commands.
901
902 DISCUSSION:
903 Half-duplex ("locked-keyboard") line-at-a-time terminals
904 for which the Go-Ahead mechanism was designed have largely
905 disappeared from the scene. It turned out to be difficult
906 to implement sending the Go-Ahead signal in many operating
907 systems, even some systems that support native half-duplex
908 terminals. The difficulty is typically that the Telnet
909 server code does not have access to information about
910 whether the user process is blocked awaiting input from
911 the Telnet connection, i.e., it cannot reliably determine
912 when to send a GA command. Therefore, most Telnet Server
913 hosts do not send GA commands.
914
915 The effect of the rules in this section is to allow either
916 end of a Telnet connection to veto the use of GA commands.
917
918 There is a class of half-duplex terminals that is still
919 commercially important: "data entry terminals," which
920 interact in a full-screen manner. However, supporting
921 data entry terminals using the Telnet protocol does not
922 require the Go Ahead signal; see <a href="#section-3.3.2">Section 3.3.2</a>.
923
924 3.2.3 Control Functions: <a href="./rfc854">RFC-854</a>, pp. 7-8
925
926 The list of Telnet commands has been extended to include EOR
927 (End-of-Record), with code 239 [TELNET:9].
928
929 Both User and Server Telnets MAY support the control functions
930 EOR, EC, EL, and Break, and MUST support AO, AYT, DM, IP, NOP,
931 SB, and SE.
932
933 A host MUST be able to receive and ignore any Telnet control
934 functions that it does not support.
935
936 DISCUSSION:
937 Note that a Server Telnet is required to support the
938 Telnet IP (Interrupt Process) function, even if the server
939 host has an equivalent in-stream function (e.g., Control-C
940 in many systems). The Telnet IP function may be stronger
941 than an in-stream interrupt command, because of the out-
942 of-band effect of TCP urgent data.
943
944 The EOR control function may be used to delimit the
945
946
947
948 <span class="grey">Internet Engineering Task Force [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
951
952
953 stream. An important application is data entry terminal
954 support (see <a href="#section-3.3.2">Section 3.3.2</a>). There was concern that since
955 EOR had not been defined in <a href="./rfc854">RFC-854</a>, a host that was not
956 prepared to correctly ignore unknown Telnet commands might
957 crash if it received an EOR. To protect such hosts, the
958 End-of-Record option [TELNET:9] was introduced; however, a
959 properly implemented Telnet program will not require this
960 protection.
961
962 3.2.4 Telnet "Synch" Signal: <a href="./rfc854">RFC-854</a>, pp. 8-10
963
964 When it receives "urgent" TCP data, a User or Server Telnet
965 MUST discard all data except Telnet commands until the DM (and
966 end of urgent) is reached.
967
968 When it sends Telnet IP (Interrupt Process), a User Telnet
969 SHOULD follow it by the Telnet "Synch" sequence, i.e., send as
970 TCP urgent data the sequence "IAC IP IAC DM". The TCP urgent
971 pointer points to the DM octet.
972
973 When it receives a Telnet IP command, a Server Telnet MAY send
974 a Telnet "Synch" sequence back to the user, to flush the output
975 stream. The choice ought to be consistent with the way the
976 server operating system behaves when a local user interrupts a
977 process.
978
979 When it receives a Telnet AO command, a Server Telnet MUST send
980 a Telnet "Synch" sequence back to the user, to flush the output
981 stream.
982
983 A User Telnet SHOULD have the capability of flushing output
984 when it sends a Telnet IP; see also <a href="#section-3.4.5">Section 3.4.5</a>.
985
986 DISCUSSION:
987 There are three possible ways for a User Telnet to flush
988 the stream of server output data:
989
990 (1) Send AO after IP.
991
992 This will cause the server host to send a "flush-
993 buffered-output" signal to its operating system.
994 However, the AO may not take effect locally, i.e.,
995 stop terminal output at the User Telnet end, until
996 the Server Telnet has received and processed the AO
997 and has sent back a "Synch".
998
999 (2) Send DO TIMING-MARK [TELNET:7] after IP, and discard
1000 all output locally until a WILL/WONT TIMING-MARK is
1001
1002
1003
1004 <span class="grey">Internet Engineering Task Force [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1007
1008
1009 received from the Server Telnet.
1010
1011 Since the DO TIMING-MARK will be processed after the
1012 IP at the server, the reply to it should be in the
1013 right place in the output data stream. However, the
1014 TIMING-MARK will not send a "flush buffered output"
1015 signal to the server operating system. Whether or
1016 not this is needed is dependent upon the server
1017 system.
1018
1019 (3) Do both.
1020
1021 The best method is not entirely clear, since it must
1022 accommodate a number of existing server hosts that do not
1023 follow the Telnet standards in various ways. The safest
1024 approach is probably to provide a user-controllable option
1025 to select (1), (2), or (3).
1026
1027 3.2.5 NVT Printer and Keyboard: <a href="./rfc854">RFC-854</a>, p. 11
1028
1029 In NVT mode, a Telnet SHOULD NOT send characters with the
1030 high-order bit 1, and MUST NOT send it as a parity bit.
1031 Implementations that pass the high-order bit to applications
1032 SHOULD negotiate binary mode (see <a href="#section-3.2.6">Section 3.2.6</a>).
1033
1034
1035 DISCUSSION:
1036 Implementors should be aware that a strict reading of
1037 <a href="./rfc854">RFC-854</a> allows a client or server expecting NVT ASCII to
1038 ignore characters with the high-order bit set. In
1039 general, binary mode is expected to be used for
1040 transmission of an extended (beyond 7-bit) character set
1041 with Telnet.
1042
1043 However, there exist applications that really need an 8-
1044 bit NVT mode, which is currently not defined, and these
1045 existing applications do set the high-order bit during
1046 part or all of the life of a Telnet connection. Note that
1047 binary mode is not the same as 8-bit NVT mode, since
1048 binary mode turns off end-of-line processing. For this
1049 reason, the requirements on the high-order bit are stated
1050 as SHOULD, not MUST.
1051
1052 <a href="./rfc854">RFC-854</a> defines a minimal set of properties of a "network
1053 virtual terminal" or NVT; this is not meant to preclude
1054 additional features in a real terminal. A Telnet
1055 connection is fully transparent to all 7-bit ASCII
1056 characters, including arbitrary ASCII control characters.
1057
1058
1059
1060 <span class="grey">Internet Engineering Task Force [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1063
1064
1065 For example, a terminal might support full-screen commands
1066 coded as ASCII escape sequences; a Telnet implementation
1067 would pass these sequences as uninterpreted data. Thus,
1068 an NVT should not be conceived as a terminal type of a
1069 highly-restricted device.
1070
1071 3.2.6 Telnet Command Structure: <a href="./rfc854">RFC-854</a>, p. 13
1072
1073 Since options may appear at any point in the data stream, a
1074 Telnet escape character (known as IAC, with the value 255) to
1075 be sent as data MUST be doubled.
1076
1077 3.2.7 Telnet Binary Option: <a href="./rfc856">RFC-856</a>
1078
1079 When the Binary option has been successfully negotiated,
1080 arbitrary 8-bit characters are allowed. However, the data
1081 stream MUST still be scanned for IAC characters, any embedded
1082 Telnet commands MUST be obeyed, and data bytes equal to IAC
1083 MUST be doubled. Other character processing (e.g., replacing
1084 CR by CR NUL or by CR LF) MUST NOT be done. In particular,
1085 there is no end-of-line convention (see <a href="#section-3.3.1">Section 3.3.1</a>) in
1086 binary mode.
1087
1088 DISCUSSION:
1089 The Binary option is normally negotiated in both
1090 directions, to change the Telnet connection from NVT mode
1091 to "binary mode".
1092
1093 The sequence IAC EOR can be used to delimit blocks of data
1094 within a binary-mode Telnet stream.
1095
1096 3.2.8 Telnet Terminal-Type Option: <a href="./rfc1091">RFC-1091</a>
1097
1098 The Terminal-Type option MUST use the terminal type names
1099 officially defined in the Assigned Numbers RFC [INTRO:5], when
1100 they are available for the particular terminal. However, the
1101 receiver of a Terminal-Type option MUST accept any name.
1102
1103 DISCUSSION:
1104 <a href="./rfc1091">RFC-1091</a> [TELNET:10] updates an earlier version of the
1105 Terminal-Type option defined in <a href="./rfc930">RFC-930</a>. The earlier
1106 version allowed a server host capable of supporting
1107 multiple terminal types to learn the type of a particular
1108 client's terminal, assuming that each physical terminal
1109 had an intrinsic type. However, today a "terminal" is
1110 often really a terminal emulator program running in a PC,
1111 perhaps capable of emulating a range of terminal types.
1112 Therefore, <a href="./rfc1091">RFC-1091</a> extends the specification to allow a
1113
1114
1115
1116 <span class="grey">Internet Engineering Task Force [Page 20]</span></pre>
1117 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
1118 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1119
1120
1121 more general terminal-type negotiation between User and
1122 Server Telnets.
1123
1124 3.3 SPECIFIC ISSUES
1125
1126 3.3.1 Telnet End-of-Line Convention
1127
1128 The Telnet protocol defines the sequence CR LF to mean "end-
1129 of-line". For terminal input, this corresponds to a command-
1130 completion or "end-of-line" key being pressed on a user
1131 terminal; on an ASCII terminal, this is the CR key, but it may
1132 also be labelled "Return" or "Enter".
1133
1134 When a Server Telnet receives the Telnet end-of-line sequence
1135 CR LF as input from a remote terminal, the effect MUST be the
1136 same as if the user had pressed the "end-of-line" key on a
1137 local terminal. On server hosts that use ASCII, in particular,
1138 receipt of the Telnet sequence CR LF must cause the same effect
1139 as a local user pressing the CR key on a local terminal. Thus,
1140 CR LF and CR NUL MUST have the same effect on an ASCII server
1141 host when received as input over a Telnet connection.
1142
1143 A User Telnet MUST be able to send any of the forms: CR LF, CR
1144 NUL, and LF. A User Telnet on an ASCII host SHOULD have a
1145 user-controllable mode to send either CR LF or CR NUL when the
1146 user presses the "end-of-line" key, and CR LF SHOULD be the
1147 default.
1148
1149 The Telnet end-of-line sequence CR LF MUST be used to send
1150 Telnet data that is not terminal-to-computer (e.g., for Server
1151 Telnet sending output, or the Telnet protocol incorporated
1152 another application protocol).
1153
1154 DISCUSSION:
1155 To allow interoperability between arbitrary Telnet clients
1156 and servers, the Telnet protocol defined a standard
1157 representation for a line terminator. Since the ASCII
1158 character set includes no explicit end-of-line character,
1159 systems have chosen various representations, e.g., CR, LF,
1160 and the sequence CR LF. The Telnet protocol chose the CR
1161 LF sequence as the standard for network transmission.
1162
1163 Unfortunately, the Telnet protocol specification in <a href="./rfc854">RFC-</a>
1164 <a href="./rfc854">854</a> [TELNET:1] has turned out to be somewhat ambiguous on
1165 what character(s) should be sent from client to server for
1166 the "end-of-line" key. The result has been a massive and
1167 continuing interoperability headache, made worse by
1168 various faulty implementations of both User and Server
1169
1170
1171
1172 <span class="grey">Internet Engineering Task Force [Page 21]</span></pre>
1173 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
1174 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1175
1176
1177 Telnets.
1178
1179 Although the Telnet protocol is based on a perfectly
1180 symmetric model, in a remote login session the role of the
1181 user at a terminal differs from the role of the server
1182 host. For example, <a href="./rfc854">RFC-854</a> defines the meaning of CR, LF,
1183 and CR LF as output from the server, but does not specify
1184 what the User Telnet should send when the user presses the
1185 "end-of-line" key on the terminal; this turns out to be
1186 the point at issue.
1187
1188 When a user presses the "end-of-line" key, some User
1189 Telnet implementations send CR LF, while others send CR
1190 NUL (based on a different interpretation of the same
1191 sentence in <a href="./rfc854">RFC-854</a>). These will be equivalent for a
1192 correctly-implemented ASCII server host, as discussed
1193 above. For other servers, a mode in the User Telnet is
1194 needed.
1195
1196 The existence of User Telnets that send only CR NUL when
1197 CR is pressed creates a dilemma for non-ASCII hosts: they
1198 can either treat CR NUL as equivalent to CR LF in input,
1199 thus precluding the possibility of entering a "bare" CR,
1200 or else lose complete interworking.
1201
1202 Suppose a user on host A uses Telnet to log into a server
1203 host B, and then execute B's User Telnet program to log
1204 into server host C. It is desirable for the Server/User
1205 Telnet combination on B to be as transparent as possible,
1206 i.e., to appear as if A were connected directly to C. In
1207 particular, correct implementation will make B transparent
1208 to Telnet end-of-line sequences, except that CR LF may be
1209 translated to CR NUL or vice versa.
1210
1211 IMPLEMENTATION:
1212 To understand Telnet end-of-line issues, one must have at
1213 least a general model of the relationship of Telnet to the
1214 local operating system. The Server Telnet process is
1215 typically coupled into the terminal driver software of the
1216 operating system as a pseudo-terminal. A Telnet end-of-
1217 line sequence received by the Server Telnet must have the
1218 same effect as pressing the end-of-line key on a real
1219 locally-connected terminal.
1220
1221 Operating systems that support interactive character-at-
1222 a-time applications (e.g., editors) typically have two
1223 internal modes for their terminal I/O: a formatted mode,
1224 in which local conventions for end-of-line and other
1225
1226
1227
1228 <span class="grey">Internet Engineering Task Force [Page 22]</span></pre>
1229 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
1230 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1231
1232
1233 formatting rules have been applied to the data stream, and
1234 a "raw" mode, in which the application has direct access
1235 to every character as it was entered. A Server Telnet
1236 must be implemented in such a way that these modes have
1237 the same effect for remote as for local terminals. For
1238 example, suppose a CR LF or CR NUL is received by the
1239 Server Telnet on an ASCII host. In raw mode, a CR
1240 character is passed to the application; in formatted mode,
1241 the local system's end-of-line convention is used.
1242
1243 3.3.2 Data Entry Terminals
1244
1245 DISCUSSION:
1246 In addition to the line-oriented and character-oriented
1247 ASCII terminals for which Telnet was designed, there are
1248 several families of video display terminals that are
1249 sometimes known as "data entry terminals" or DETs. The
1250 IBM 3270 family is a well-known example.
1251
1252 Two Internet protocols have been designed to support
1253 generic DETs: SUPDUP [TELNET:16, TELNET:17], and the DET
1254 option [TELNET:18, TELNET:19]. The DET option drives a
1255 data entry terminal over a Telnet connection using (sub-)
1256 negotiation. SUPDUP is a completely separate terminal
1257 protocol, which can be entered from Telnet by negotiation.
1258 Although both SUPDUP and the DET option have been used
1259 successfully in particular environments, neither has
1260 gained general acceptance or wide implementation.
1261
1262 A different approach to DET interaction has been developed
1263 for supporting the IBM 3270 family through Telnet,
1264 although the same approach would be applicable to any DET.
1265 The idea is to enter a "native DET" mode, in which the
1266 native DET input/output stream is sent as binary data.
1267 The Telnet EOR command is used to delimit logical records
1268 (e.g., "screens") within this binary stream.
1269
1270 IMPLEMENTATION:
1271 The rules for entering and leaving native DET mode are as
1272 follows:
1273
1274 o The Server uses the Terminal-Type option [TELNET:10]
1275 to learn that the client is a DET.
1276
1277 o It is conventional, but not required, that both ends
1278 negotiate the EOR option [TELNET:9].
1279
1280 o Both ends negotiate the Binary option [TELNET:3] to
1281
1282
1283
1284 <span class="grey">Internet Engineering Task Force [Page 23]</span></pre>
1285 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
1286 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1287
1288
1289 enter native DET mode.
1290
1291 o When either end negotiates out of binary mode, the
1292 other end does too, and the mode then reverts to
1293 normal NVT.
1294
1295
1296 3.3.3 Option Requirements
1297
1298 Every Telnet implementation MUST support the Binary option
1299 [TELNET:3] and the Suppress Go Ahead option [TELNET:5], and
1300 SHOULD support the Echo [TELNET:4], Status [TELNET:6], End-of-
1301 Record [TELNET:9], and Extended Options List [TELNET:8]
1302 options.
1303
1304 A User or Server Telnet SHOULD support the Window Size Option
1305 [TELNET:12] if the local operating system provides the
1306 corresponding capability.
1307
1308 DISCUSSION:
1309 Note that the End-of-Record option only signifies that a
1310 Telnet can receive a Telnet EOR without crashing;
1311 therefore, every Telnet ought to be willing to accept
1312 negotiation of the End-of-Record option. See also the
1313 discussion in <a href="#section-3.2.3">Section 3.2.3</a>.
1314
1315 3.3.4 Option Initiation
1316
1317 When the Telnet protocol is used in a client/server situation,
1318 the server SHOULD initiate negotiation of the terminal
1319 interaction mode it expects.
1320
1321 DISCUSSION:
1322 The Telnet protocol was defined to be perfectly
1323 symmetrical, but its application is generally asymmetric.
1324 Remote login has been known to fail because NEITHER side
1325 initiated negotiation of the required non-default terminal
1326 modes. It is generally the server that determines the
1327 preferred mode, so the server needs to initiate the
1328 negotiation; since the negotiation is symmetric, the user
1329 can also initiate it.
1330
1331 A client (User Telnet) SHOULD provide a means for users to
1332 enable and disable the initiation of option negotiation.
1333
1334 DISCUSSION:
1335 A user sometimes needs to connect to an application
1336 service (e.g., FTP or SMTP) that uses Telnet for its
1337
1338
1339
1340 <span class="grey">Internet Engineering Task Force [Page 24]</span></pre>
1341 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
1342 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1343
1344
1345 control stream but does not support Telnet options. User
1346 Telnet may be used for this purpose if initiation of
1347 option negotiation is disabled.
1348
1349 3.3.5 Telnet Linemode Option
1350
1351 DISCUSSION:
1352 An important new Telnet option, LINEMODE [TELNET:12], has
1353 been proposed. The LINEMODE option provides a standard
1354 way for a User Telnet and a Server Telnet to agree that
1355 the client rather than the server will perform terminal
1356 character processing. When the client has prepared a
1357 complete line of text, it will send it to the server in
1358 (usually) one TCP packet. This option will greatly
1359 decrease the packet cost of Telnet sessions and will also
1360 give much better user response over congested or long-
1361 delay networks.
1362
1363 The LINEMODE option allows dynamic switching between local
1364 and remote character processing. For example, the Telnet
1365 connection will automatically negotiate into single-
1366 character mode while a full screen editor is running, and
1367 then return to linemode when the editor is finished.
1368
1369 We expect that when this RFC is released, hosts should
1370 implement the client side of this option, and may
1371 implement the server side of this option. To properly
1372 implement the server side, the server needs to be able to
1373 tell the local system not to do any input character
1374 processing, but to remember its current terminal state and
1375 notify the Server Telnet process whenever the state
1376 changes. This will allow password echoing and full screen
1377 editors to be handled properly, for example.
1378
1379 3.4 TELNET/USER INTERFACE
1380
1381 3.4.1 Character Set Transparency
1382
1383 User Telnet implementations SHOULD be able to send or receive
1384 any 7-bit ASCII character. Where possible, any special
1385 character interpretations by the user host's operating system
1386 SHOULD be bypassed so that these characters can conveniently be
1387 sent and received on the connection.
1388
1389 Some character value MUST be reserved as "escape to command
1390 mode"; conventionally, doubling this character allows it to be
1391 entered as data. The specific character used SHOULD be user
1392 selectable.
1393
1394
1395
1396 <span class="grey">Internet Engineering Task Force [Page 25]</span></pre>
1397 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
1398 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1399
1400
1401 On binary-mode connections, a User Telnet program MAY provide
1402 an escape mechanism for entering arbitrary 8-bit values, if the
1403 host operating system doesn't allow them to be entered directly
1404 from the keyboard.
1405
1406 IMPLEMENTATION:
1407 The transparency issues are less pressing on servers, but
1408 implementors should take care in dealing with issues like:
1409 masking off parity bits (sent by an older, non-conforming
1410 client) before they reach programs that expect only NVT
1411 ASCII, and properly handling programs that request 8-bit
1412 data streams.
1413
1414 3.4.2 Telnet Commands
1415
1416 A User Telnet program MUST provide a user the capability of
1417 entering any of the Telnet control functions IP, AO, or AYT,
1418 and SHOULD provide the capability of entering EC, EL, and
1419 Break.
1420
1421 3.4.3 TCP Connection Errors
1422
1423 A User Telnet program SHOULD report to the user any TCP errors
1424 that are reported by the transport layer (see "TCP/Application
1425 Layer Interface" section in [INTRO:1]).
1426
1427 3.4.4 Non-Default Telnet Contact Port
1428
1429 A User Telnet program SHOULD allow the user to optionally
1430 specify a non-standard contact port number at the Server Telnet
1431 host.
1432
1433 3.4.5 Flushing Output
1434
1435 A User Telnet program SHOULD provide the user the ability to
1436 specify whether or not output should be flushed when an IP is
1437 sent; see <a href="#section-3.2.4">Section 3.2.4</a>.
1438
1439 For any output flushing scheme that causes the User Telnet to
1440 flush output locally until a Telnet signal is received from the
1441 Server, there SHOULD be a way for the user to manually restore
1442 normal output, in case the Server fails to send the expected
1443 signal.
1444
1445
1446
1447
1448
1449
1450
1451
1452 <span class="grey">Internet Engineering Task Force [Page 26]</span></pre>
1453 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-27" ></span>
1454 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1455
1456
1457 3.5. TELNET REQUIREMENTS SUMMARY
1458
1459
1460 | | | | |S| |
1461 | | | | |H| |F
1462 | | | | |O|M|o
1463 | | |S| |U|U|o
1464 | | |H| |L|S|t
1465 | |M|O| |D|T|n
1466 | |U|U|M| | |o
1467 | |S|L|A|N|N|t
1468 | |T|D|Y|O|O|t
1469 FEATURE |SECTION | | | |T|T|e
1470 -------------------------------------------------|--------|-|-|-|-|-|--
1471 | | | | | | |
1472 Option Negotiation |3.2.1 |x| | | | |
1473 Avoid negotiation loops |3.2.1 |x| | | | |
1474 Refuse unsupported options |3.2.1 |x| | | | |
1475 Negotiation OK anytime on connection |3.2.1 | |x| | | |
1476 Default to NVT |3.2.1 |x| | | | |
1477 Send official name in Term-Type option |3.2.8 |x| | | | |
1478 Accept any name in Term-Type option |3.2.8 |x| | | | |
1479 Implement Binary, Suppress-GA options |3.3.3 |x| | | | |
1480 Echo, Status, EOL, Ext-Opt-List options |3.3.3 | |x| | | |
1481 Implement Window-Size option if appropriate |3.3.3 | |x| | | |
1482 Server initiate mode negotiations |3.3.4 | |x| | | |
1483 User can enable/disable init negotiations |3.3.4 | |x| | | |
1484 | | | | | | |
1485 Go-Aheads | | | | | | |
1486 Non-GA server negotiate SUPPRESS-GA option |3.2.2 |x| | | | |
1487 User or Server accept SUPPRESS-GA option |3.2.2 |x| | | | |
1488 User Telnet ignore GA's |3.2.2 | | |x| | |
1489 | | | | | | |
1490 Control Functions | | | | | | |
1491 Support SE NOP DM IP AO AYT SB |3.2.3 |x| | | | |
1492 Support EOR EC EL Break |3.2.3 | | |x| | |
1493 Ignore unsupported control functions |3.2.3 |x| | | | |
1494 User, Server discard urgent data up to DM |3.2.4 |x| | | | |
1495 User Telnet send "Synch" after IP, AO, AYT |3.2.4 | |x| | | |
1496 Server Telnet reply Synch to IP |3.2.4 | | |x| | |
1497 Server Telnet reply Synch to AO |3.2.4 |x| | | | |
1498 User Telnet can flush output when send IP |3.2.4 | |x| | | |
1499 | | | | | | |
1500 Encoding | | | | | | |
1501 Send high-order bit in NVT mode |3.2.5 | | | |x| |
1502 Send high-order bit as parity bit |3.2.5 | | | | |x|
1503 Negot. BINARY if pass high-ord. bit to applic |3.2.5 | |x| | | |
1504 Always double IAC data byte |3.2.6 |x| | | | |
1505
1506
1507
1508 <span class="grey">Internet Engineering Task Force [Page 27]</span></pre>
1509 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-28" ></span>
1510 <span class="grey"><a href="./rfc1123">RFC1123</a> REMOTE LOGIN -- TELNET October 1989</span>
1511
1512
1513 Double IAC data byte in binary mode |3.2.7 |x| | | | |
1514 Obey Telnet cmds in binary mode |3.2.7 |x| | | | |
1515 End-of-line, CR NUL in binary mode |3.2.7 | | | | |x|
1516 | | | | | | |
1517 End-of-Line | | | | | | |
1518 EOL at Server same as local end-of-line |3.3.1 |x| | | | |
1519 ASCII Server accept CR LF or CR NUL for EOL |3.3.1 |x| | | | |
1520 User Telnet able to send CR LF, CR NUL, or LF |3.3.1 |x| | | | |
1521 ASCII user able to select CR LF/CR NUL |3.3.1 | |x| | | |
1522 User Telnet default mode is CR LF |3.3.1 | |x| | | |
1523 Non-interactive uses CR LF for EOL |3.3.1 |x| | | | |
1524 | | | | | | |
1525 User Telnet interface | | | | | | |
1526 Input &amp; output all 7-bit characters |3.4.1 | |x| | | |
1527 Bypass local op sys interpretation |3.4.1 | |x| | | |
1528 Escape character |3.4.1 |x| | | | |
1529 User-settable escape character |3.4.1 | |x| | | |
1530 Escape to enter 8-bit values |3.4.1 | | |x| | |
1531 Can input IP, AO, AYT |3.4.2 |x| | | | |
1532 Can input EC, EL, Break |3.4.2 | |x| | | |
1533 Report TCP connection errors to user |3.4.3 | |x| | | |
1534 Optional non-default contact port |3.4.4 | |x| | | |
1535 Can spec: output flushed when IP sent |3.4.5 | |x| | | |
1536 Can manually restore output mode |3.4.5 | |x| | | |
1537 | | | | | | |
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564 <span class="grey">Internet Engineering Task Force [Page 28]</span></pre>
1565 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-29" ></span>
1566 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1567
1568
1569 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. FILE TRANSFER</span>
1570
1571 4.1 FILE TRANSFER PROTOCOL -- FTP
1572
1573 4.1.1 INTRODUCTION
1574
1575 The File Transfer Protocol FTP is the primary Internet standard
1576 for file transfer. The current specification is contained in
1577 <a href="./rfc959">RFC-959</a> [FTP:1].
1578
1579 FTP uses separate simultaneous TCP connections for control and
1580 for data transfer. The FTP protocol includes many features,
1581 some of which are not commonly implemented. However, for every
1582 feature in FTP, there exists at least one implementation. The
1583 minimum implementation defined in <a href="./rfc959">RFC-959</a> was too small, so a
1584 somewhat larger minimum implementation is defined here.
1585
1586 Internet users have been unnecessarily burdened for years by
1587 deficient FTP implementations. Protocol implementors have
1588 suffered from the erroneous opinion that implementing FTP ought
1589 to be a small and trivial task. This is wrong, because FTP has
1590 a user interface, because it has to deal (correctly) with the
1591 whole variety of communication and operating system errors that
1592 may occur, and because it has to handle the great diversity of
1593 real file systems in the world.
1594
1595 4.1.2. PROTOCOL WALK-THROUGH
1596
1597 4.1.2.1 LOCAL Type: <a href="./rfc959#section-3.1.1.4">RFC-959 Section&nbsp;3.1.1.4</a>
1598
1599 An FTP program MUST support TYPE I ("IMAGE" or binary type)
1600 as well as TYPE L 8 ("LOCAL" type with logical byte size 8).
1601 A machine whose memory is organized into m-bit words, where
1602 m is not a multiple of 8, MAY also support TYPE L m.
1603
1604 DISCUSSION:
1605 The command "TYPE L 8" is often required to transfer
1606 binary data between a machine whose memory is organized
1607 into (e.g.) 36-bit words and a machine with an 8-bit
1608 byte organization. For an 8-bit byte machine, TYPE L 8
1609 is equivalent to IMAGE.
1610
1611 "TYPE L m" is sometimes specified to the FTP programs
1612 on two m-bit word machines to ensure the correct
1613 transfer of a native-mode binary file from one machine
1614 to the other. However, this command should have the
1615 same effect on these machines as "TYPE I".
1616
1617
1618
1619
1620 <span class="grey">Internet Engineering Task Force [Page 29]</span></pre>
1621 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-30" ></span>
1622 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1623
1624
1625 4.1.2.2 Telnet Format Control: <a href="./rfc959#section-3.1.1.5.2">RFC-959 Section&nbsp;3.1.1.5.2</a>
1626
1627 A host that makes no distinction between TYPE N and TYPE T
1628 SHOULD implement TYPE T to be identical to TYPE N.
1629
1630 DISCUSSION:
1631 This provision should ease interoperation with hosts
1632 that do make this distinction.
1633
1634 Many hosts represent text files internally as strings
1635 of ASCII characters, using the embedded ASCII format
1636 effector characters (LF, BS, FF, ...) to control the
1637 format when a file is printed. For such hosts, there
1638 is no distinction between "print" files and other
1639 files. However, systems that use record structured
1640 files typically need a special format for printable
1641 files (e.g., ASA carriage control). For the latter
1642 hosts, FTP allows a choice of TYPE N or TYPE T.
1643
1644 4.1.2.3 Page Structure: <a href="./rfc959#section-3.1.2.3">RFC-959 Section&nbsp;3.1.2.3</a> and <a href="#appendix-I">Appendix I</a>
1645
1646 Implementation of page structure is NOT RECOMMENDED in
1647 general. However, if a host system does need to implement
1648 FTP for "random access" or "holey" files, it MUST use the
1649 defined page structure format rather than define a new
1650 private FTP format.
1651
1652 4.1.2.4 Data Structure Transformations: <a href="./rfc959#section-3.1.2">RFC-959 Section&nbsp;3.1.2</a>
1653
1654 An FTP transformation between record-structure and file-
1655 structure SHOULD be invertible, to the extent possible while
1656 making the result useful on the target host.
1657
1658 DISCUSSION:
1659 <a href="./rfc959">RFC-959</a> required strict invertibility between record-
1660 structure and file-structure, but in practice,
1661 efficiency and convenience often preclude it.
1662 Therefore, the requirement is being relaxed. There are
1663 two different objectives for transferring a file:
1664 processing it on the target host, or just storage. For
1665 storage, strict invertibility is important. For
1666 processing, the file created on the target host needs
1667 to be in the format expected by application programs on
1668 that host.
1669
1670 As an example of the conflict, imagine a record-
1671 oriented operating system that requires some data files
1672 to have exactly 80 bytes in each record. While STORing
1673
1674
1675
1676 <span class="grey">Internet Engineering Task Force [Page 30]</span></pre>
1677 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-31" ></span>
1678 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1679
1680
1681 a file on such a host, an FTP Server must be able to
1682 pad each line or record to 80 bytes; a later retrieval
1683 of such a file cannot be strictly invertible.
1684
1685 4.1.2.5 Data Connection Management: <a href="./rfc959#section-3.3">RFC-959 Section&nbsp;3.3</a>
1686
1687 A User-FTP that uses STREAM mode SHOULD send a PORT command
1688 to assign a non-default data port before each transfer
1689 command is issued.
1690
1691 DISCUSSION:
1692 This is required because of the long delay after a TCP
1693 connection is closed until its socket pair can be
1694 reused, to allow multiple transfers during a single FTP
1695 session. Sending a port command can avoided if a
1696 transfer mode other than stream is used, by leaving the
1697 data transfer connection open between transfers.
1698
1699 4.1.2.6 PASV Command: <a href="./rfc959#section-4.1.2">RFC-959 Section&nbsp;4.1.2</a>
1700
1701 A server-FTP MUST implement the PASV command.
1702
1703 If multiple third-party transfers are to be executed during
1704 the same session, a new PASV command MUST be issued before
1705 each transfer command, to obtain a unique port pair.
1706
1707 IMPLEMENTATION:
1708 The format of the 227 reply to a PASV command is not
1709 well standardized. In particular, an FTP client cannot
1710 assume that the parentheses shown on page 40 of <a href="./rfc959">RFC-959</a>
1711 will be present (and in fact, Figure 3 on page 43 omits
1712 them). Therefore, a User-FTP program that interprets
1713 the PASV reply must scan the reply for the first digit
1714 of the host and port numbers.
1715
1716 Note that the host number h1,h2,h3,h4 is the IP address
1717 of the server host that is sending the reply, and that
1718 p1,p2 is a non-default data transfer port that PASV has
1719 assigned.
1720
1721 4.1.2.7 LIST and NLST Commands: <a href="./rfc959#section-4.1.3">RFC-959 Section&nbsp;4.1.3</a>
1722
1723 The data returned by an NLST command MUST contain only a
1724 simple list of legal pathnames, such that the server can use
1725 them directly as the arguments of subsequent data transfer
1726 commands for the individual files.
1727
1728 The data returned by a LIST or NLST command SHOULD use an
1729
1730
1731
1732 <span class="grey">Internet Engineering Task Force [Page 31]</span></pre>
1733 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-32" ></span>
1734 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1735
1736
1737 implied TYPE AN, unless the current type is EBCDIC, in which
1738 case an implied TYPE EN SHOULD be used.
1739
1740 DISCUSSION:
1741 Many FTP clients support macro-commands that will get
1742 or put files matching a wildcard specification, using
1743 NLST to obtain a list of pathnames. The expansion of
1744 "multiple-put" is local to the client, but "multiple-
1745 get" requires cooperation by the server.
1746
1747 The implied type for LIST and NLST is designed to
1748 provide compatibility with existing User-FTPs, and in
1749 particular with multiple-get commands.
1750
1751 4.1.2.8 SITE Command: <a href="./rfc959#section-4.1.3">RFC-959 Section&nbsp;4.1.3</a>
1752
1753 A Server-FTP SHOULD use the SITE command for non-standard
1754 features, rather than invent new private commands or
1755 unstandardized extensions to existing commands.
1756
1757 4.1.2.9 STOU Command: <a href="./rfc959#section-4.1.3">RFC-959 Section&nbsp;4.1.3</a>
1758
1759 The STOU command stores into a uniquely named file. When it
1760 receives an STOU command, a Server-FTP MUST return the
1761 actual file name in the "125 Transfer Starting" or the "150
1762 Opening Data Connection" message that precedes the transfer
1763 (the 250 reply code mentioned in <a href="./rfc959">RFC-959</a> is incorrect). The
1764 exact format of these messages is hereby defined to be as
1765 follows:
1766
1767 125 FILE: pppp
1768 150 FILE: pppp
1769
1770 where pppp represents the unique pathname of the file that
1771 will be written.
1772
1773 4.1.2.10 Telnet End-of-line Code: <a href="./rfc959">RFC-959</a>, Page 34
1774
1775 Implementors MUST NOT assume any correspondence between READ
1776 boundaries on the control connection and the Telnet EOL
1777 sequences (CR LF).
1778
1779 DISCUSSION:
1780 Thus, a server-FTP (or User-FTP) must continue reading
1781 characters from the control connection until a complete
1782 Telnet EOL sequence is encountered, before processing
1783 the command (or response, respectively). Conversely, a
1784 single READ from the control connection may include
1785
1786
1787
1788 <span class="grey">Internet Engineering Task Force [Page 32]</span></pre>
1789 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-33" ></span>
1790 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1791
1792
1793 more than one FTP command.
1794
1795 4.1.2.11 FTP Replies: <a href="./rfc959#section-4.2">RFC-959 Section&nbsp;4.2</a>, Page 35
1796
1797 A Server-FTP MUST send only correctly formatted replies on
1798 the control connection. Note that <a href="./rfc959">RFC-959</a> (unlike earlier
1799 versions of the FTP spec) contains no provision for a
1800 "spontaneous" reply message.
1801
1802 A Server-FTP SHOULD use the reply codes defined in <a href="./rfc959">RFC-959</a>
1803 whenever they apply. However, a server-FTP MAY use a
1804 different reply code when needed, as long as the general
1805 rules of <a href="#section-4.2">Section 4.2</a> are followed. When the implementor has
1806 a choice between a 4xx and 5xx reply code, a Server-FTP
1807 SHOULD send a 4xx (temporary failure) code when there is any
1808 reasonable possibility that a failed FTP will succeed a few
1809 hours later.
1810
1811 A User-FTP SHOULD generally use only the highest-order digit
1812 of a 3-digit reply code for making a procedural decision, to
1813 prevent difficulties when a Server-FTP uses non-standard
1814 reply codes.
1815
1816 A User-FTP MUST be able to handle multi-line replies. If
1817 the implementation imposes a limit on the number of lines
1818 and if this limit is exceeded, the User-FTP MUST recover,
1819 e.g., by ignoring the excess lines until the end of the
1820 multi-line reply is reached.
1821
1822 A User-FTP SHOULD NOT interpret a 421 reply code ("Service
1823 not available, closing control connection") specially, but
1824 SHOULD detect closing of the control connection by the
1825 server.
1826
1827 DISCUSSION:
1828 Server implementations that fail to strictly follow the
1829 reply rules often cause FTP user programs to hang.
1830 Note that <a href="./rfc959">RFC-959</a> resolved ambiguities in the reply
1831 rules found in earlier FTP specifications and must be
1832 followed.
1833
1834 It is important to choose FTP reply codes that properly
1835 distinguish between temporary and permanent failures,
1836 to allow the successful use of file transfer client
1837 daemons. These programs depend on the reply codes to
1838 decide whether or not to retry a failed transfer; using
1839 a permanent failure code (5xx) for a temporary error
1840 will cause these programs to give up unnecessarily.
1841
1842
1843
1844 <span class="grey">Internet Engineering Task Force [Page 33]</span></pre>
1845 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-34" ></span>
1846 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1847
1848
1849 When the meaning of a reply matches exactly the text
1850 shown in <a href="./rfc959">RFC-959</a>, uniformity will be enhanced by using
1851 the <a href="./rfc959">RFC-959</a> text verbatim. However, a Server-FTP
1852 implementor is encouraged to choose reply text that
1853 conveys specific system-dependent information, when
1854 appropriate.
1855
1856 4.1.2.12 Connections: <a href="./rfc959#section-5.2">RFC-959 Section&nbsp;5.2</a>
1857
1858 The words "and the port used" in the second paragraph of
1859 this section of <a href="./rfc959">RFC-959</a> are erroneous (historical), and they
1860 should be ignored.
1861
1862 On a multihomed server host, the default data transfer port
1863 (L-1) MUST be associated with the same local IP address as
1864 the corresponding control connection to port L.
1865
1866 A user-FTP MUST NOT send any Telnet controls other than
1867 SYNCH and IP on an FTP control connection. In particular, it
1868 MUST NOT attempt to negotiate Telnet options on the control
1869 connection. However, a server-FTP MUST be capable of
1870 accepting and refusing Telnet negotiations (i.e., sending
1871 DONT/WONT).
1872
1873 DISCUSSION:
1874 Although the RFC says: "Server- and User- processes
1875 should follow the conventions for the Telnet
1876 protocol...[on the control connection]", it is not the
1877 intent that Telnet option negotiation is to be
1878 employed.
1879
1880 4.1.2.13 Minimum Implementation; <a href="./rfc959#section-5.1">RFC-959 Section&nbsp;5.1</a>
1881
1882 The following commands and options MUST be supported by
1883 every server-FTP and user-FTP, except in cases where the
1884 underlying file system or operating system does not allow or
1885 support a particular command.
1886
1887 Type: ASCII Non-print, IMAGE, LOCAL 8
1888 Mode: Stream
1889 Structure: File, Record*
1890 Commands:
1891 USER, PASS, ACCT,
1892 PORT, PASV,
1893 TYPE, MODE, STRU,
1894 RETR, STOR, APPE,
1895 RNFR, RNTO, DELE,
1896 CWD, CDUP, RMD, MKD, PWD,
1897
1898
1899
1900 <span class="grey">Internet Engineering Task Force [Page 34]</span></pre>
1901 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-35" ></span>
1902 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1903
1904
1905 LIST, NLST,
1906 SYST, STAT,
1907 HELP, NOOP, QUIT.
1908
1909 *Record structure is REQUIRED only for hosts whose file
1910 systems support record structure.
1911
1912 DISCUSSION:
1913 Vendors are encouraged to implement a larger subset of
1914 the protocol. For example, there are important
1915 robustness features in the protocol (e.g., Restart,
1916 ABOR, block mode) that would be an aid to some Internet
1917 users but are not widely implemented.
1918
1919 A host that does not have record structures in its file
1920 system may still accept files with STRU R, recording
1921 the byte stream literally.
1922
1923 4.1.3 SPECIFIC ISSUES
1924
1925 4.1.3.1 Non-standard Command Verbs
1926
1927 FTP allows "experimental" commands, whose names begin with
1928 "X". If these commands are subsequently adopted as
1929 standards, there may still be existing implementations using
1930 the "X" form. At present, this is true for the directory
1931 commands:
1932
1933 <a href="./rfc959">RFC-959</a> "Experimental"
1934
1935 MKD XMKD
1936 RMD XRMD
1937 PWD XPWD
1938 CDUP XCUP
1939 CWD XCWD
1940
1941 All FTP implementations SHOULD recognize both forms of these
1942 commands, by simply equating them with extra entries in the
1943 command lookup table.
1944
1945 IMPLEMENTATION:
1946 A User-FTP can access a server that supports only the
1947 "X" forms by implementing a mode switch, or
1948 automatically using the following procedure: if the
1949 <a href="./rfc959">RFC-959</a> form of one of the above commands is rejected
1950 with a 500 or 502 response code, then try the
1951 experimental form; any other response would be passed
1952 to the user.
1953
1954
1955
1956 <span class="grey">Internet Engineering Task Force [Page 35]</span></pre>
1957 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-36" ></span>
1958 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
1959
1960
1961 4.1.3.2 Idle Timeout
1962
1963 A Server-FTP process SHOULD have an idle timeout, which will
1964 terminate the process and close the control connection if
1965 the server is inactive (i.e., no command or data transfer in
1966 progress) for a long period of time. The idle timeout time
1967 SHOULD be configurable, and the default should be at least 5
1968 minutes.
1969
1970 A client FTP process ("User-PI" in <a href="./rfc959">RFC-959</a>) will need
1971 timeouts on responses only if it is invoked from a program.
1972
1973 DISCUSSION:
1974 Without a timeout, a Server-FTP process may be left
1975 pending indefinitely if the corresponding client
1976 crashes without closing the control connection.
1977
1978 4.1.3.3 Concurrency of Data and Control
1979
1980 DISCUSSION:
1981 The intent of the designers of FTP was that a user
1982 should be able to send a STAT command at any time while
1983 data transfer was in progress and that the server-FTP
1984 would reply immediately with status -- e.g., the number
1985 of bytes transferred so far. Similarly, an ABOR
1986 command should be possible at any time during a data
1987 transfer.
1988
1989 Unfortunately, some small-machine operating systems
1990 make such concurrent programming difficult, and some
1991 other implementers seek minimal solutions, so some FTP
1992 implementations do not allow concurrent use of the data
1993 and control connections. Even such a minimal server
1994 must be prepared to accept and defer a STAT or ABOR
1995 command that arrives during data transfer.
1996
1997 4.1.3.4 FTP Restart Mechanism
1998
1999 The description of the 110 reply on pp. 40-41 of <a href="./rfc959">RFC-959</a> is
2000 incorrect; the correct description is as follows. A restart
2001 reply message, sent over the control connection from the
2002 receiving FTP to the User-FTP, has the format:
2003
2004 110 MARK ssss = rrrr
2005
2006 Here:
2007
2008 * ssss is a text string that appeared in a Restart Marker
2009
2010
2011
2012 <span class="grey">Internet Engineering Task Force [Page 36]</span></pre>
2013 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-37" ></span>
2014 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2015
2016
2017 in the data stream and encodes a position in the
2018 sender's file system;
2019
2020 * rrrr encodes the corresponding position in the
2021 receiver's file system.
2022
2023 The encoding, which is specific to a particular file system
2024 and network implementation, is always generated and
2025 interpreted by the same system, either sender or receiver.
2026
2027 When an FTP that implements restart receives a Restart
2028 Marker in the data stream, it SHOULD force the data to that
2029 point to be written to stable storage before encoding the
2030 corresponding position rrrr. An FTP sending Restart Markers
2031 MUST NOT assume that 110 replies will be returned
2032 synchronously with the data, i.e., it must not await a 110
2033 reply before sending more data.
2034
2035 Two new reply codes are hereby defined for errors
2036 encountered in restarting a transfer:
2037
2038 554 Requested action not taken: invalid REST parameter.
2039
2040 A 554 reply may result from a FTP service command that
2041 follows a REST command. The reply indicates that the
2042 existing file at the Server-FTP cannot be repositioned
2043 as specified in the REST.
2044
2045 555 Requested action not taken: type or stru mismatch.
2046
2047 A 555 reply may result from an APPE command or from any
2048 FTP service command following a REST command. The
2049 reply indicates that there is some mismatch between the
2050 current transfer parameters (type and stru) and the
2051 attributes of the existing file.
2052
2053 DISCUSSION:
2054 Note that the FTP Restart mechanism requires that Block
2055 or Compressed mode be used for data transfer, to allow
2056 the Restart Markers to be included within the data
2057 stream. The frequency of Restart Markers can be low.
2058
2059 Restart Markers mark a place in the data stream, but
2060 the receiver may be performing some transformation on
2061 the data as it is stored into stable storage. In
2062 general, the receiver's encoding must include any state
2063 information necessary to restart this transformation at
2064 any point of the FTP data stream. For example, in TYPE
2065
2066
2067
2068 <span class="grey">Internet Engineering Task Force [Page 37]</span></pre>
2069 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-38" ></span>
2070 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2071
2072
2073 A transfers, some receiver hosts transform CR LF
2074 sequences into a single LF character on disk. If a
2075 Restart Marker happens to fall between CR and LF, the
2076 receiver must encode in rrrr that the transfer must be
2077 restarted in a "CR has been seen and discarded" state.
2078
2079 Note that the Restart Marker is required to be encoded
2080 as a string of printable ASCII characters, regardless
2081 of the type of the data.
2082
2083 <a href="./rfc959">RFC-959</a> says that restart information is to be returned
2084 "to the user". This should not be taken literally. In
2085 general, the User-FTP should save the restart
2086 information (ssss,rrrr) in stable storage, e.g., append
2087 it to a restart control file. An empty restart control
2088 file should be created when the transfer first starts
2089 and deleted automatically when the transfer completes
2090 successfully. It is suggested that this file have a
2091 name derived in an easily-identifiable manner from the
2092 name of the file being transferred and the remote host
2093 name; this is analogous to the means used by many text
2094 editors for naming "backup" files.
2095
2096 There are three cases for FTP restart.
2097
2098 (1) User-to-Server Transfer
2099
2100 The User-FTP puts Restart Markers &lt;ssss&gt; at
2101 convenient places in the data stream. When the
2102 Server-FTP receives a Marker, it writes all prior
2103 data to disk, encodes its file system position and
2104 transformation state as rrrr, and returns a "110
2105 MARK ssss = rrrr" reply over the control
2106 connection. The User-FTP appends the pair
2107 (ssss,rrrr) to its restart control file.
2108
2109 To restart the transfer, the User-FTP fetches the
2110 last (ssss,rrrr) pair from the restart control
2111 file, repositions its local file system and
2112 transformation state using ssss, and sends the
2113 command "REST rrrr" to the Server-FTP.
2114
2115 (2) Server-to-User Transfer
2116
2117 The Server-FTP puts Restart Markers &lt;ssss&gt; at
2118 convenient places in the data stream. When the
2119 User-FTP receives a Marker, it writes all prior
2120 data to disk, encodes its file system position and
2121
2122
2123
2124 <span class="grey">Internet Engineering Task Force [Page 38]</span></pre>
2125 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-39" ></span>
2126 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2127
2128
2129 transformation state as rrrr, and appends the pair
2130 (rrrr,ssss) to its restart control file.
2131
2132 To restart the transfer, the User-FTP fetches the
2133 last (rrrr,ssss) pair from the restart control
2134 file, repositions its local file system and
2135 transformation state using rrrr, and sends the
2136 command "REST ssss" to the Server-FTP.
2137
2138 (3) Server-to-Server ("Third-Party") Transfer
2139
2140 The sending Server-FTP puts Restart Markers &lt;ssss&gt;
2141 at convenient places in the data stream. When it
2142 receives a Marker, the receiving Server-FTP writes
2143 all prior data to disk, encodes its file system
2144 position and transformation state as rrrr, and
2145 sends a "110 MARK ssss = rrrr" reply over the
2146 control connection to the User. The User-FTP
2147 appends the pair (ssss,rrrr) to its restart
2148 control file.
2149
2150 To restart the transfer, the User-FTP fetches the
2151 last (ssss,rrrr) pair from the restart control
2152 file, sends "REST ssss" to the sending Server-FTP,
2153 and sends "REST rrrr" to the receiving Server-FTP.
2154
2155
2156 4.1.4 FTP/USER INTERFACE
2157
2158 This section discusses the user interface for a User-FTP
2159 program.
2160
2161 4.1.4.1 Pathname Specification
2162
2163 Since FTP is intended for use in a heterogeneous
2164 environment, User-FTP implementations MUST support remote
2165 pathnames as arbitrary character strings, so that their form
2166 and content are not limited by the conventions of the local
2167 operating system.
2168
2169 DISCUSSION:
2170 In particular, remote pathnames can be of arbitrary
2171 length, and all the printing ASCII characters as well
2172 as space (0x20) must be allowed. <a href="./rfc959">RFC-959</a> allows a
2173 pathname to contain any 7-bit ASCII character except CR
2174 or LF.
2175
2176
2177
2178
2179
2180 <span class="grey">Internet Engineering Task Force [Page 39]</span></pre>
2181 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-40" ></span>
2182 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2183
2184
2185 4.1.4.2 "QUOTE" Command
2186
2187 A User-FTP program MUST implement a "QUOTE" command that
2188 will pass an arbitrary character string to the server and
2189 display all resulting response messages to the user.
2190
2191 To make the "QUOTE" command useful, a User-FTP SHOULD send
2192 transfer control commands to the server as the user enters
2193 them, rather than saving all the commands and sending them
2194 to the server only when a data transfer is started.
2195
2196 DISCUSSION:
2197 The "QUOTE" command is essential to allow the user to
2198 access servers that require system-specific commands
2199 (e.g., SITE or ALLO), or to invoke new or optional
2200 features that are not implemented by the User-FTP. For
2201 example, "QUOTE" may be used to specify "TYPE A T" to
2202 send a print file to hosts that require the
2203 distinction, even if the User-FTP does not recognize
2204 that TYPE.
2205
2206 4.1.4.3 Displaying Replies to User
2207
2208 A User-FTP SHOULD display to the user the full text of all
2209 error reply messages it receives. It SHOULD have a
2210 "verbose" mode in which all commands it sends and the full
2211 text and reply codes it receives are displayed, for
2212 diagnosis of problems.
2213
2214 4.1.4.4 Maintaining Synchronization
2215
2216 The state machine in a User-FTP SHOULD be forgiving of
2217 missing and unexpected reply messages, in order to maintain
2218 command synchronization with the server.
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236 <span class="grey">Internet Engineering Task Force [Page 40]</span></pre>
2237 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-41" ></span>
2238 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2239
2240
2241 4.1.5 FTP REQUIREMENTS SUMMARY
2242
2243 | | | | |S| |
2244 | | | | |H| |F
2245 | | | | |O|M|o
2246 | | |S| |U|U|o
2247 | | |H| |L|S|t
2248 | |M|O| |D|T|n
2249 | |U|U|M| | |o
2250 | |S|L|A|N|N|t
2251 | |T|D|Y|O|O|t
2252 FEATURE |SECTION | | | |T|T|e
2253 -------------------------------------------|---------------|-|-|-|-|-|--
2254 Implement TYPE T if same as TYPE N |4.1.2.2 | |x| | | |
2255 File/Record transform invertible if poss. |4.1.2.4 | |x| | | |
2256 User-FTP send PORT cmd for stream mode |4.1.2.5 | |x| | | |
2257 Server-FTP implement PASV |4.1.2.6 |x| | | | |
2258 PASV is per-transfer |4.1.2.6 |x| | | | |
2259 NLST reply usable in RETR cmds |4.1.2.7 |x| | | | |
2260 Implied type for LIST and NLST |4.1.2.7 | |x| | | |
2261 SITE cmd for non-standard features |4.1.2.8 | |x| | | |
2262 STOU cmd return pathname as specified |4.1.2.9 |x| | | | |
2263 Use TCP READ boundaries on control conn. |4.1.2.10 | | | | |x|
2264 | | | | | | |
2265 Server-FTP send only correct reply format |4.1.2.11 |x| | | | |
2266 Server-FTP use defined reply code if poss. |4.1.2.11 | |x| | | |
2267 New reply code following <a href="#section-4.2">Section 4.2</a> |4.1.2.11 | | |x| | |
2268 User-FTP use only high digit of reply |4.1.2.11 | |x| | | |
2269 User-FTP handle multi-line reply lines |4.1.2.11 |x| | | | |
2270 User-FTP handle 421 reply specially |4.1.2.11 | | | |x| |
2271 | | | | | | |
2272 Default data port same IP addr as ctl conn |4.1.2.12 |x| | | | |
2273 User-FTP send Telnet cmds exc. SYNCH, IP |4.1.2.12 | | | | |x|
2274 User-FTP negotiate Telnet options |4.1.2.12 | | | | |x|
2275 Server-FTP handle Telnet options |4.1.2.12 |x| | | | |
2276 Handle "Experimental" directory cmds |4.1.3.1 | |x| | | |
2277 Idle timeout in server-FTP |4.1.3.2 | |x| | | |
2278 Configurable idle timeout |4.1.3.2 | |x| | | |
2279 Receiver checkpoint data at Restart Marker |4.1.3.4 | |x| | | |
2280 Sender assume 110 replies are synchronous |4.1.3.4 | | | | |x|
2281 | | | | | | |
2282 Support TYPE: | | | | | | |
2283 ASCII - Non-Print (AN) |4.1.2.13 |x| | | | |
2284 ASCII - Telnet (AT) -- if same as AN |4.1.2.2 | |x| | | |
2285 ASCII - Carriage Control (AC) |959 3.1.1.5.2 | | |x| | |
2286 EBCDIC - (any form) |959 3.1.1.2 | | |x| | |
2287 IMAGE |4.1.2.1 |x| | | | |
2288 LOCAL 8 |4.1.2.1 |x| | | | |
2289
2290
2291
2292 <span class="grey">Internet Engineering Task Force [Page 41]</span></pre>
2293 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-42" ></span>
2294 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2295
2296
2297 LOCAL m |4.1.2.1 | | |x| | |2
2298 | | | | | | |
2299 Support MODE: | | | | | | |
2300 Stream |4.1.2.13 |x| | | | |
2301 Block |959 3.4.2 | | |x| | |
2302 | | | | | | |
2303 Support STRUCTURE: | | | | | | |
2304 File |4.1.2.13 |x| | | | |
2305 Record |4.1.2.13 |x| | | | |3
2306 Page |4.1.2.3 | | | |x| |
2307 | | | | | | |
2308 Support commands: | | | | | | |
2309 USER |4.1.2.13 |x| | | | |
2310 PASS |4.1.2.13 |x| | | | |
2311 ACCT |4.1.2.13 |x| | | | |
2312 CWD |4.1.2.13 |x| | | | |
2313 CDUP |4.1.2.13 |x| | | | |
2314 SMNT |959 5.3.1 | | |x| | |
2315 REIN |959 5.3.1 | | |x| | |
2316 QUIT |4.1.2.13 |x| | | | |
2317 | | | | | | |
2318 PORT |4.1.2.13 |x| | | | |
2319 PASV |4.1.2.6 |x| | | | |
2320 TYPE |4.1.2.13 |x| | | | |1
2321 STRU |4.1.2.13 |x| | | | |1
2322 MODE |4.1.2.13 |x| | | | |1
2323 | | | | | | |
2324 RETR |4.1.2.13 |x| | | | |
2325 STOR |4.1.2.13 |x| | | | |
2326 STOU |959 5.3.1 | | |x| | |
2327 APPE |4.1.2.13 |x| | | | |
2328 ALLO |959 5.3.1 | | |x| | |
2329 REST |959 5.3.1 | | |x| | |
2330 RNFR |4.1.2.13 |x| | | | |
2331 RNTO |4.1.2.13 |x| | | | |
2332 ABOR |959 5.3.1 | | |x| | |
2333 DELE |4.1.2.13 |x| | | | |
2334 RMD |4.1.2.13 |x| | | | |
2335 MKD |4.1.2.13 |x| | | | |
2336 PWD |4.1.2.13 |x| | | | |
2337 LIST |4.1.2.13 |x| | | | |
2338 NLST |4.1.2.13 |x| | | | |
2339 SITE |4.1.2.8 | | |x| | |
2340 STAT |4.1.2.13 |x| | | | |
2341 SYST |4.1.2.13 |x| | | | |
2342 HELP |4.1.2.13 |x| | | | |
2343 NOOP |4.1.2.13 |x| | | | |
2344 | | | | | | |
2345
2346
2347
2348 <span class="grey">Internet Engineering Task Force [Page 42]</span></pre>
2349 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-43" ></span>
2350 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- FTP October 1989</span>
2351
2352
2353 User Interface: | | | | | | |
2354 Arbitrary pathnames |4.1.4.1 |x| | | | |
2355 Implement "QUOTE" command |4.1.4.2 |x| | | | |
2356 Transfer control commands immediately |4.1.4.2 | |x| | | |
2357 Display error messages to user |4.1.4.3 | |x| | | |
2358 Verbose mode |4.1.4.3 | |x| | | |
2359 Maintain synchronization with server |4.1.4.4 | |x| | | |
2360
2361 Footnotes:
2362
2363 (1) For the values shown earlier.
2364
2365 (2) Here m is number of bits in a memory word.
2366
2367 (3) Required for host with record-structured file system, optional
2368 otherwise.
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404 <span class="grey">Internet Engineering Task Force [Page 43]</span></pre>
2405 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-44" ></span>
2406 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- TFTP October 1989</span>
2407
2408
2409 4.2 TRIVIAL FILE TRANSFER PROTOCOL -- TFTP
2410
2411 4.2.1 INTRODUCTION
2412
2413 The Trivial File Transfer Protocol TFTP is defined in <a href="./rfc783">RFC-783</a>
2414 [TFTP:1].
2415
2416 TFTP provides its own reliable delivery with UDP as its
2417 transport protocol, using a simple stop-and-wait acknowledgment
2418 system. Since TFTP has an effective window of only one 512
2419 octet segment, it can provide good performance only over paths
2420 that have a small delay*bandwidth product. The TFTP file
2421 interface is very simple, providing no access control or
2422 security.
2423
2424 TFTP's most important application is bootstrapping a host over
2425 a local network, since it is simple and small enough to be
2426 easily implemented in EPROM [BOOT:1, BOOT:2]. Vendors are
2427 urged to support TFTP for booting.
2428
2429 4.2.2 PROTOCOL WALK-THROUGH
2430
2431 The TFTP specification [TFTP:1] is written in an open style,
2432 and does not fully specify many parts of the protocol.
2433
2434 4.2.2.1 Transfer Modes: <a href="./rfc783">RFC-783</a>, Page 3
2435
2436 The transfer mode "mail" SHOULD NOT be supported.
2437
2438 4.2.2.2 UDP Header: <a href="./rfc783">RFC-783</a>, Page 17
2439
2440 The Length field of a UDP header is incorrectly defined; it
2441 includes the UDP header length (8).
2442
2443 4.2.3 SPECIFIC ISSUES
2444
2445 4.2.3.1 Sorcerer's Apprentice Syndrome
2446
2447 There is a serious bug, known as the "Sorcerer's Apprentice
2448 Syndrome," in the protocol specification. While it does not
2449 cause incorrect operation of the transfer (the file will
2450 always be transferred correctly if the transfer completes),
2451 this bug may cause excessive retransmission, which may cause
2452 the transfer to time out.
2453
2454 Implementations MUST contain the fix for this problem: the
2455 sender (i.e., the side originating the DATA packets) must
2456 never resend the current DATA packet on receipt of a
2457
2458
2459
2460 <span class="grey">Internet Engineering Task Force [Page 44]</span></pre>
2461 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-45" ></span>
2462 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- TFTP October 1989</span>
2463
2464
2465 duplicate ACK.
2466
2467 DISCUSSION:
2468 The bug is caused by the protocol rule that either
2469 side, on receiving an old duplicate datagram, may
2470 resend the current datagram. If a packet is delayed in
2471 the network but later successfully delivered after
2472 either side has timed out and retransmitted a packet, a
2473 duplicate copy of the response may be generated. If
2474 the other side responds to this duplicate with a
2475 duplicate of its own, then every datagram will be sent
2476 in duplicate for the remainder of the transfer (unless
2477 a datagram is lost, breaking the repetition). Worse
2478 yet, since the delay is often caused by congestion,
2479 this duplicate transmission will usually causes more
2480 congestion, leading to more delayed packets, etc.
2481
2482 The following example may help to clarify this problem.
2483
2484 TFTP A TFTP B
2485
2486 (1) Receive ACK X-1
2487 Send DATA X
2488 (2) Receive DATA X
2489 Send ACK X
2490 (ACK X is delayed in network,
2491 and A times out):
2492 (3) Retransmit DATA X
2493
2494 (4) Receive DATA X again
2495 Send ACK X again
2496 (5) Receive (delayed) ACK X
2497 Send DATA X+1
2498 (6) Receive DATA X+1
2499 Send ACK X+1
2500 (7) Receive ACK X again
2501 Send DATA X+1 again
2502 (8) Receive DATA X+1 again
2503 Send ACK X+1 again
2504 (9) Receive ACK X+1
2505 Send DATA X+2
2506 (10) Receive DATA X+2
2507 Send ACK X+3
2508 (11) Receive ACK X+1 again
2509 Send DATA X+2 again
2510 (12) Receive DATA X+2 again
2511 Send ACK X+3 again
2512
2513
2514
2515
2516 <span class="grey">Internet Engineering Task Force [Page 45]</span></pre>
2517 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-46" ></span>
2518 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- TFTP October 1989</span>
2519
2520
2521 Notice that once the delayed ACK arrives, the protocol
2522 settles down to duplicate all further packets
2523 (sequences 5-8 and 9-12). The problem is caused not by
2524 either side timing out, but by both sides
2525 retransmitting the current packet when they receive a
2526 duplicate.
2527
2528 The fix is to break the retransmission loop, as
2529 indicated above. This is analogous to the behavior of
2530 TCP. It is then possible to remove the retransmission
2531 timer on the receiver, since the resent ACK will never
2532 cause any action; this is a useful simplification where
2533 TFTP is used in a bootstrap program. It is OK to allow
2534 the timer to remain, and it may be helpful if the
2535 retransmitted ACK replaces one that was genuinely lost
2536 in the network. The sender still requires a retransmit
2537 timer, of course.
2538
2539 4.2.3.2 Timeout Algorithms
2540
2541 A TFTP implementation MUST use an adaptive timeout.
2542
2543 IMPLEMENTATION:
2544 TCP retransmission algorithms provide a useful base to
2545 work from. At least an exponential backoff of
2546 retransmission timeout is necessary.
2547
2548 4.2.3.3 Extensions
2549
2550 A variety of non-standard extensions have been made to TFTP,
2551 including additional transfer modes and a secure operation
2552 mode (with passwords). None of these have been
2553 standardized.
2554
2555 4.2.3.4 Access Control
2556
2557 A server TFTP implementation SHOULD include some
2558 configurable access control over what pathnames are allowed
2559 in TFTP operations.
2560
2561 4.2.3.5 Broadcast Request
2562
2563 A TFTP request directed to a broadcast address SHOULD be
2564 silently ignored.
2565
2566 DISCUSSION:
2567 Due to the weak access control capability of TFTP,
2568 directed broadcasts of TFTP requests to random networks
2569
2570
2571
2572 <span class="grey">Internet Engineering Task Force [Page 46]</span></pre>
2573 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-47" ></span>
2574 <span class="grey"><a href="./rfc1123">RFC1123</a> FILE TRANSFER -- TFTP October 1989</span>
2575
2576
2577 could create a significant security hole.
2578
2579 4.2.4 TFTP REQUIREMENTS SUMMARY
2580
2581 | | | | |S| |
2582 | | | | |H| |F
2583 | | | | |O|M|o
2584 | | |S| |U|U|o
2585 | | |H| |L|S|t
2586 | |M|O| |D|T|n
2587 | |U|U|M| | |o
2588 | |S|L|A|N|N|t
2589 | |T|D|Y|O|O|t
2590 FEATURE |SECTION | | | |T|T|e
2591 -------------------------------------------------|--------|-|-|-|-|-|--
2592 Fix Sorcerer's Apprentice Syndrome |4.2.3.1 |x| | | | |
2593 Transfer modes: | | | | | | |
2594 netascii |<a href="./rfc783">RFC-783</a> |x| | | | |
2595 octet |<a href="./rfc783">RFC-783</a> |x| | | | |
2596 mail |4.2.2.1 | | | |x| |
2597 extensions |4.2.3.3 | | |x| | |
2598 Use adaptive timeout |4.2.3.2 |x| | | | |
2599 Configurable access control |4.2.3.4 | |x| | | |
2600 Silently ignore broadcast request |4.2.3.5 | |x| | | |
2601 -------------------------------------------------|--------|-|-|-|-|-|--
2602 -------------------------------------------------|--------|-|-|-|-|-|--
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628 <span class="grey">Internet Engineering Task Force [Page 47]</span></pre>
2629 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-48" ></span>
2630 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2631
2632
2633 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. ELECTRONIC MAIL -- SMTP and <a href="./rfc822">RFC-822</a></span>
2634
2635 5.1 INTRODUCTION
2636
2637 In the TCP/IP protocol suite, electronic mail in a format
2638 specified in <a href="./rfc822">RFC-822</a> [SMTP:2] is transmitted using the Simple Mail
2639 Transfer Protocol (SMTP) defined in <a href="./rfc821">RFC-821</a> [SMTP:1].
2640
2641 While SMTP has remained unchanged over the years, the Internet
2642 community has made several changes in the way SMTP is used. In
2643 particular, the conversion to the Domain Name System (DNS) has
2644 caused changes in address formats and in mail routing. In this
2645 section, we assume familiarity with the concepts and terminology
2646 of the DNS, whose requirements are given in <a href="#section-6.1">Section 6.1</a>.
2647
2648 <a href="./rfc822">RFC-822</a> specifies the Internet standard format for electronic mail
2649 messages. <a href="./rfc822">RFC-822</a> supercedes an older standard, <a href="./rfc733">RFC-733</a>, that may
2650 still be in use in a few places, although it is obsolete. The two
2651 formats are sometimes referred to simply by number ("822" and
2652 "733").
2653
2654 <a href="./rfc822">RFC-822</a> is used in some non-Internet mail environments with
2655 different mail transfer protocols than SMTP, and SMTP has also
2656 been adapted for use in some non-Internet environments. Note that
2657 this document presents the rules for the use of SMTP and <a href="./rfc822">RFC-822</a>
2658 for the Internet environment only; other mail environments that
2659 use these protocols may be expected to have their own rules.
2660
2661 5.2 PROTOCOL WALK-THROUGH
2662
2663 This section covers both <a href="./rfc821">RFC-821</a> and <a href="./rfc822">RFC-822</a>.
2664
2665 The SMTP specification in <a href="./rfc821">RFC-821</a> is clear and contains numerous
2666 examples, so implementors should not find it difficult to
2667 understand. This section simply updates or annotates portions of
2668 <a href="./rfc821">RFC-821</a> to conform with current usage.
2669
2670 <a href="./rfc822">RFC-822</a> is a long and dense document, defining a rich syntax.
2671 Unfortunately, incomplete or defective implementations of <a href="./rfc822">RFC-822</a>
2672 are common. In fact, nearly all of the many formats of <a href="./rfc822">RFC-822</a>
2673 are actually used, so an implementation generally needs to
2674 recognize and correctly interpret all of the <a href="./rfc822">RFC-822</a> syntax.
2675
2676 5.2.1 The SMTP Model: <a href="./rfc821#section-2">RFC-821 Section&nbsp;2</a>
2677
2678 DISCUSSION:
2679 Mail is sent by a series of request/response transactions
2680 between a client, the "sender-SMTP," and a server, the
2681
2682
2683
2684 <span class="grey">Internet Engineering Task Force [Page 48]</span></pre>
2685 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-49" ></span>
2686 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2687
2688
2689 "receiver-SMTP". These transactions pass (1) the message
2690 proper, which is composed of header and body, and (2) SMTP
2691 source and destination addresses, referred to as the
2692 "envelope".
2693
2694 The SMTP programs are analogous to Message Transfer Agents
2695 (MTAs) of X.400. There will be another level of protocol
2696 software, closer to the end user, that is responsible for
2697 composing and analyzing <a href="./rfc822">RFC-822</a> message headers; this
2698 component is known as the "User Agent" in X.400, and we
2699 use that term in this document. There is a clear logical
2700 distinction between the User Agent and the SMTP
2701 implementation, since they operate on different levels of
2702 protocol. Note, however, that this distinction is may not
2703 be exactly reflected the structure of typical
2704 implementations of Internet mail. Often there is a
2705 program known as the "mailer" that implements SMTP and
2706 also some of the User Agent functions; the rest of the
2707 User Agent functions are included in a user interface used
2708 for entering and reading mail.
2709
2710 The SMTP envelope is constructed at the originating site,
2711 typically by the User Agent when the message is first
2712 queued for the Sender-SMTP program. The envelope
2713 addresses may be derived from information in the message
2714 header, supplied by the user interface (e.g., to implement
2715 a bcc: request), or derived from local configuration
2716 information (e.g., expansion of a mailing list). The SMTP
2717 envelope cannot in general be re-derived from the header
2718 at a later stage in message delivery, so the envelope is
2719 transmitted separately from the message itself using the
2720 MAIL and RCPT commands of SMTP.
2721
2722 The text of <a href="./rfc821">RFC-821</a> suggests that mail is to be delivered
2723 to an individual user at a host. With the advent of the
2724 domain system and of mail routing using mail-exchange (MX)
2725 resource records, implementors should now think of
2726 delivering mail to a user at a domain, which may or may
2727 not be a particular host. This DOES NOT change the fact
2728 that SMTP is a host-to-host mail exchange protocol.
2729
2730 5.2.2 Canonicalization: <a href="./rfc821#section-3.1">RFC-821 Section&nbsp;3.1</a>
2731
2732 The domain names that a Sender-SMTP sends in MAIL and RCPT
2733 commands MUST have been "canonicalized," i.e., they must be
2734 fully-qualified principal names or domain literals, not
2735 nicknames or domain abbreviations. A canonicalized name either
2736 identifies a host directly or is an MX name; it cannot be a
2737
2738
2739
2740 <span class="grey">Internet Engineering Task Force [Page 49]</span></pre>
2741 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-50" ></span>
2742 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2743
2744
2745 CNAME.
2746
2747 5.2.3 VRFY and EXPN Commands: <a href="./rfc821#section-3.3">RFC-821 Section&nbsp;3.3</a>
2748
2749 A receiver-SMTP MUST implement VRFY and SHOULD implement EXPN
2750 (this requirement overrides <a href="./rfc821">RFC-821</a>). However, there MAY be
2751 configuration information to disable VRFY and EXPN in a
2752 particular installation; this might even allow EXPN to be
2753 disabled for selected lists.
2754
2755 A new reply code is defined for the VRFY command:
2756
2757 252 Cannot VRFY user (e.g., info is not local), but will
2758 take message for this user and attempt delivery.
2759
2760 DISCUSSION:
2761 SMTP users and administrators make regular use of these
2762 commands for diagnosing mail delivery problems. With the
2763 increasing use of multi-level mailing list expansion
2764 (sometimes more than two levels), EXPN has been
2765 increasingly important for diagnosing inadvertent mail
2766 loops. On the other hand, some feel that EXPN represents
2767 a significant privacy, and perhaps even a security,
2768 exposure.
2769
2770 5.2.4 SEND, SOML, and SAML Commands: <a href="./rfc821#section-3.4">RFC-821 Section&nbsp;3.4</a>
2771
2772 An SMTP MAY implement the commands to send a message to a
2773 user's terminal: SEND, SOML, and SAML.
2774
2775 DISCUSSION:
2776 It has been suggested that the use of mail relaying
2777 through an MX record is inconsistent with the intent of
2778 SEND to deliver a message immediately and directly to a
2779 user's terminal. However, an SMTP receiver that is unable
2780 to write directly to the user terminal can return a "251
2781 User Not Local" reply to the RCPT following a SEND, to
2782 inform the originator of possibly deferred delivery.
2783
2784 5.2.5 HELO Command: <a href="./rfc821#section-3.5">RFC-821 Section&nbsp;3.5</a>
2785
2786 The sender-SMTP MUST ensure that the &lt;domain&gt; parameter in a
2787 HELO command is a valid principal host domain name for the
2788 client host. As a result, the receiver-SMTP will not have to
2789 perform MX resolution on this name in order to validate the
2790 HELO parameter.
2791
2792 The HELO receiver MAY verify that the HELO parameter really
2793
2794
2795
2796 <span class="grey">Internet Engineering Task Force [Page 50]</span></pre>
2797 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-51" ></span>
2798 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2799
2800
2801 corresponds to the IP address of the sender. However, the
2802 receiver MUST NOT refuse to accept a message, even if the
2803 sender's HELO command fails verification.
2804
2805 DISCUSSION:
2806 Verifying the HELO parameter requires a domain name lookup
2807 and may therefore take considerable time. An alternative
2808 tool for tracking bogus mail sources is suggested below
2809 (see "DATA Command").
2810
2811 Note also that the HELO argument is still required to have
2812 valid &lt;domain&gt; syntax, since it will appear in a Received:
2813 line; otherwise, a 501 error is to be sent.
2814
2815 IMPLEMENTATION:
2816 When HELO parameter validation fails, a suggested
2817 procedure is to insert a note about the unknown
2818 authenticity of the sender into the message header (e.g.,
2819 in the "Received:" line).
2820
2821 5.2.6 Mail Relay: <a href="./rfc821#section-3.6">RFC-821 Section&nbsp;3.6</a>
2822
2823 We distinguish three types of mail (store-and-) forwarding:
2824
2825 (1) A simple forwarder or "mail exchanger" forwards a message
2826 using private knowledge about the recipient; see <a href="./rfc821#section-3.2">section</a>
2827 <a href="./rfc821#section-3.2">3.2 of RFC-821</a>.
2828
2829 (2) An SMTP mail "relay" forwards a message within an SMTP
2830 mail environment as the result of an explicit source route
2831 (as defined in <a href="./rfc821#section-3.6">section&nbsp;3.6 of RFC-821</a>). The SMTP relay
2832 function uses the "@...:" form of source route from <a href="./rfc822">RFC-</a>
2833 <a href="./rfc822">822</a> (see <a href="#section-5.2.19">Section 5.2.19</a> below).
2834
2835 (3) A mail "gateway" passes a message between different
2836 environments. The rules for mail gateways are discussed
2837 below in <a href="#section-5.3.7">Section 5.3.7</a>.
2838
2839 An Internet host that is forwarding a message but is not a
2840 gateway to a different mail environment (i.e., it falls under
2841 (1) or (2)) SHOULD NOT alter any existing header fields,
2842 although the host will add an appropriate Received: line as
2843 required in <a href="#section-5.2.8">Section 5.2.8</a>.
2844
2845 A Sender-SMTP SHOULD NOT send a RCPT TO: command containing an
2846 explicit source route using the "@...:" address form. Thus,
2847 the relay function defined in section 3.6 of <a href="./rfc821">RFC-821</a> should
2848 not be used.
2849
2850
2851
2852 <span class="grey">Internet Engineering Task Force [Page 51]</span></pre>
2853 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-52" ></span>
2854 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2855
2856
2857 DISCUSSION:
2858 The intent is to discourage all source routing and to
2859 abolish explicit source routing for mail delivery within
2860 the Internet environment. Source-routing is unnecessary;
2861 the simple target address "user@domain" should always
2862 suffice. This is the result of an explicit architectural
2863 decision to use universal naming rather than source
2864 routing for mail. Thus, SMTP provides end-to-end
2865 connectivity, and the DNS provides globally-unique,
2866 location-independent names. MX records handle the major
2867 case where source routing might otherwise be needed.
2868
2869 A receiver-SMTP MUST accept the explicit source route syntax in
2870 the envelope, but it MAY implement the relay function as
2871 defined in <a href="./rfc821#section-3.6">section&nbsp;3.6 of RFC-821</a>. If it does not implement
2872 the relay function, it SHOULD attempt to deliver the message
2873 directly to the host to the right of the right-most "@" sign.
2874
2875 DISCUSSION:
2876 For example, suppose a host that does not implement the
2877 relay function receives a message with the SMTP command:
2878 "RCPT TO:&lt;@ALPHA,@BETA:joe@GAMMA&gt;", where ALPHA, BETA, and
2879 GAMMA represent domain names. Rather than immediately
2880 refusing the message with a 550 error reply as suggested
2881 on page 20 of <a href="./rfc821">RFC-821</a>, the host should try to forward the
2882 message to GAMMA directly, using: "RCPT TO:&lt;joe@GAMMA&gt;".
2883 Since this host does not support relaying, it is not
2884 required to update the reverse path.
2885
2886 Some have suggested that source routing may be needed
2887 occasionally for manually routing mail around failures;
2888 however, the reality and importance of this need is
2889 controversial. The use of explicit SMTP mail relaying for
2890 this purpose is discouraged, and in fact it may not be
2891 successful, as many host systems do not support it. Some
2892 have used the "%-hack" (see <a href="#section-5.2.16">Section 5.2.16</a>) for this
2893 purpose.
2894
2895 5.2.7 RCPT Command: <a href="./rfc821#section-4.1.1">RFC-821 Section&nbsp;4.1.1</a>
2896
2897 A host that supports a receiver-SMTP MUST support the reserved
2898 mailbox "Postmaster".
2899
2900 The receiver-SMTP MAY verify RCPT parameters as they arrive;
2901 however, RCPT responses MUST NOT be delayed beyond a reasonable
2902 time (see <a href="#section-5.3.2">Section 5.3.2</a>).
2903
2904 Therefore, a "250 OK" response to a RCPT does not necessarily
2905
2906
2907
2908 <span class="grey">Internet Engineering Task Force [Page 52]</span></pre>
2909 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-53" ></span>
2910 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2911
2912
2913 imply that the delivery address(es) are valid. Errors found
2914 after message acceptance will be reported by mailing a
2915 notification message to an appropriate address (see <a href="#section-5.3.3">Section</a>
2916 <a href="#section-5.3.3">5.3.3</a>).
2917
2918 DISCUSSION:
2919 The set of conditions under which a RCPT parameter can be
2920 validated immediately is an engineering design choice.
2921 Reporting destination mailbox errors to the Sender-SMTP
2922 before mail is transferred is generally desirable to save
2923 time and network bandwidth, but this advantage is lost if
2924 RCPT verification is lengthy.
2925
2926 For example, the receiver can verify immediately any
2927 simple local reference, such as a single locally-
2928 registered mailbox. On the other hand, the "reasonable
2929 time" limitation generally implies deferring verification
2930 of a mailing list until after the message has been
2931 transferred and accepted, since verifying a large mailing
2932 list can take a very long time. An implementation might
2933 or might not choose to defer validation of addresses that
2934 are non-local and therefore require a DNS lookup. If a
2935 DNS lookup is performed but a soft domain system error
2936 (e.g., timeout) occurs, validity must be assumed.
2937
2938 5.2.8 DATA Command: <a href="./rfc821#section-4.1.1">RFC-821 Section&nbsp;4.1.1</a>
2939
2940 Every receiver-SMTP (not just one that "accepts a message for
2941 relaying or for final delivery" [SMTP:1]) MUST insert a
2942 "Received:" line at the beginning of a message. In this line,
2943 called a "time stamp line" in <a href="./rfc821">RFC-821</a>:
2944
2945 * The FROM field SHOULD contain both (1) the name of the
2946 source host as presented in the HELO command and (2) a
2947 domain literal containing the IP address of the source,
2948 determined from the TCP connection.
2949
2950 * The ID field MAY contain an "@" as suggested in <a href="./rfc822">RFC-822</a>,
2951 but this is not required.
2952
2953 * The FOR field MAY contain a list of &lt;path&gt; entries when
2954 multiple RCPT commands have been given.
2955
2956
2957 An Internet mail program MUST NOT change a Received: line that
2958 was previously added to the message header.
2959
2960
2961
2962
2963
2964 <span class="grey">Internet Engineering Task Force [Page 53]</span></pre>
2965 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-54" ></span>
2966 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
2967
2968
2969 DISCUSSION:
2970 Including both the source host and the IP source address
2971 in the Received: line may provide enough information for
2972 tracking illicit mail sources and eliminate a need to
2973 explicitly verify the HELO parameter.
2974
2975 Received: lines are primarily intended for humans tracing
2976 mail routes, primarily of diagnosis of faults. See also
2977 the discussion under 5.3.7.
2978
2979 When the receiver-SMTP makes "final delivery" of a message,
2980 then it MUST pass the MAIL FROM: address from the SMTP envelope
2981 with the message, for use if an error notification message must
2982 be sent later (see <a href="#section-5.3.3">Section 5.3.3</a>). There is an analogous
2983 requirement when gatewaying from the Internet into a different
2984 mail environment; see <a href="#section-5.3.7">Section 5.3.7</a>.
2985
2986 DISCUSSION:
2987 Note that the final reply to the DATA command depends only
2988 upon the successful transfer and storage of the message.
2989 Any problem with the destination address(es) must either
2990 (1) have been reported in an SMTP error reply to the RCPT
2991 command(s), or (2) be reported in a later error message
2992 mailed to the originator.
2993
2994 IMPLEMENTATION:
2995 The MAIL FROM: information may be passed as a parameter or
2996 in a Return-Path: line inserted at the beginning of the
2997 message.
2998
2999 5.2.9 Command Syntax: <a href="./rfc821#section-4.1.2">RFC-821 Section&nbsp;4.1.2</a>
3000
3001 The syntax shown in <a href="./rfc821">RFC-821</a> for the MAIL FROM: command omits
3002 the case of an empty path: "MAIL FROM: &lt;&gt;" (see <a href="./rfc821">RFC-821</a> Page
3003 15). An empty reverse path MUST be supported.
3004
3005 5.2.10 SMTP Replies: <a href="./rfc821#section-4.2">RFC-821 Section&nbsp;4.2</a>
3006
3007 A receiver-SMTP SHOULD send only the reply codes listed in
3008 <a href="./rfc821#section-4.2.2">section&nbsp;4.2.2 of RFC-821</a> or in this document. A receiver-SMTP
3009 SHOULD use the text shown in examples in <a href="./rfc821">RFC-821</a> whenever
3010 appropriate.
3011
3012 A sender-SMTP MUST determine its actions only by the reply
3013 code, not by the text (except for 251 and 551 replies); any
3014 text, including no text at all, must be acceptable. The space
3015 (blank) following the reply code is considered part of the
3016 text. Whenever possible, a sender-SMTP SHOULD test only the
3017
3018
3019
3020 <span class="grey">Internet Engineering Task Force [Page 54]</span></pre>
3021 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-55" ></span>
3022 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3023
3024
3025 first digit of the reply code, as specified in <a href="./rfc821#appendix-E">Appendix&nbsp;E of
3026 RFC-821</a>.
3027
3028 DISCUSSION:
3029 Interoperability problems have arisen with SMTP systems
3030 using reply codes that are not listed explicitly in <a href="./rfc821">RFC-</a>
3031 <a href="./rfc821">821</a> <a href="#section-4.3">Section 4.3</a> but are legal according to the theory of
3032 reply codes explained in <a href="#appendix-E">Appendix E</a>.
3033
3034 5.2.11 Transparency: <a href="./rfc821#section-4.5.2">RFC-821 Section&nbsp;4.5.2</a>
3035
3036 Implementors MUST be sure that their mail systems always add
3037 and delete periods to ensure message transparency.
3038
3039 5.2.12 WKS Use in MX Processing: <a href="./rfc974">RFC-974</a>, p. 5
3040
3041 <a href="./rfc974">RFC-974</a> [SMTP:3] recommended that the domain system be queried
3042 for WKS ("Well-Known Service") records, to verify that each
3043 proposed mail target does support SMTP. Later experience has
3044 shown that WKS is not widely supported, so the WKS step in MX
3045 processing SHOULD NOT be used.
3046
3047 The following are notes on <a href="./rfc822">RFC-822</a>, organized by section of that
3048 document.
3049
3050 5.2.13 <a href="./rfc822">RFC-822</a> Message Specification: <a href="./rfc822#section-4">RFC-822 Section&nbsp;4</a>
3051
3052 The syntax shown for the Return-path line omits the possibility
3053 of a null return path, which is used to prevent looping of
3054 error notifications (see <a href="#section-5.3.3">Section 5.3.3</a>). The complete syntax
3055 is:
3056
3057 return = "Return-path" ":" route-addr
3058 / "Return-path" ":" "&lt;" "&gt;"
3059
3060 The set of optional header fields is hereby expanded to include
3061 the Content-Type field defined in <a href="./rfc1049">RFC-1049</a> [SMTP:7]. This
3062 field "allows mail reading systems to automatically identify
3063 the type of a structured message body and to process it for
3064 display accordingly". [SMTP:7] A User Agent MAY support this
3065 field.
3066
3067 5.2.14 <a href="./rfc822">RFC-822</a> Date and Time Specification: <a href="./rfc822#section-5">RFC-822 Section&nbsp;5</a>
3068
3069 The syntax for the date is hereby changed to:
3070
3071 date = 1*2DIGIT month 2*4DIGIT
3072
3073
3074
3075
3076 <span class="grey">Internet Engineering Task Force [Page 55]</span></pre>
3077 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-56" ></span>
3078 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3079
3080
3081 All mail software SHOULD use 4-digit years in dates, to ease
3082 the transition to the next century.
3083
3084 There is a strong trend towards the use of numeric timezone
3085 indicators, and implementations SHOULD use numeric timezones
3086 instead of timezone names. However, all implementations MUST
3087 accept either notation. If timezone names are used, they MUST
3088 be exactly as defined in <a href="./rfc822">RFC-822</a>.
3089
3090 The military time zones are specified incorrectly in <a href="./rfc822">RFC-822</a>:
3091 they count the wrong way from UT (the signs are reversed). As
3092 a result, military time zones in <a href="./rfc822">RFC-822</a> headers carry no
3093 information.
3094
3095 Finally, note that there is a typo in the definition of "zone"
3096 in the syntax summary of <a href="#appendix-D">appendix D</a>; the correct definition
3097 occurs in <a href="./rfc822#section-3">Section&nbsp;3 of RFC-822</a>.
3098
3099 5.2.15 <a href="./rfc822">RFC-822</a> Syntax Change: <a href="./rfc822#section-6.1">RFC-822 Section&nbsp;6.1</a>
3100
3101 The syntactic definition of "mailbox" in <a href="./rfc822">RFC-822</a> is hereby
3102 changed to:
3103
3104 mailbox = addr-spec ; simple address
3105 / [phrase] route-addr ; name &amp; addr-spec
3106
3107 That is, the phrase preceding a route address is now OPTIONAL.
3108 This change makes the following header field legal, for
3109 example:
3110
3111 From: &lt;[email protected]&gt;
3112
3113 5.2.16 <a href="./rfc822">RFC-822</a> Local-part: <a href="./rfc822#section-6.2">RFC-822 Section&nbsp;6.2</a>
3114
3115 The basic mailbox address specification has the form: "local-
3116 part@domain". Here "local-part", sometimes called the "left-
3117 hand side" of the address, is domain-dependent.
3118
3119 A host that is forwarding the message but is not the
3120 destination host implied by the right-hand side "domain" MUST
3121 NOT interpret or modify the "local-part" of the address.
3122
3123 When mail is to be gatewayed from the Internet mail environment
3124 into a foreign mail environment (see <a href="#section-5.3.7">Section 5.3.7</a>), routing
3125 information for that foreign environment MAY be embedded within
3126 the "local-part" of the address. The gateway will then
3127 interpret this local part appropriately for the foreign mail
3128 environment.
3129
3130
3131
3132 <span class="grey">Internet Engineering Task Force [Page 56]</span></pre>
3133 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-57" ></span>
3134 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3135
3136
3137 DISCUSSION:
3138 Although source routes are discouraged within the Internet
3139 (see <a href="#section-5.2.6">Section 5.2.6</a>), there are non-Internet mail
3140 environments whose delivery mechanisms do depend upon
3141 source routes. Source routes for extra-Internet
3142 environments can generally be buried in the "local-part"
3143 of the address (see <a href="#section-5.2.16">Section 5.2.16</a>) while mail traverses
3144 the Internet. When the mail reaches the appropriate
3145 Internet mail gateway, the gateway will interpret the
3146 local-part and build the necessary address or route for
3147 the target mail environment.
3148
3149 For example, an Internet host might send mail to:
3150 "a!b!c!user@gateway-domain". The complex local part
3151 "a!b!c!user" would be uninterpreted within the Internet
3152 domain, but could be parsed and understood by the
3153 specified mail gateway.
3154
3155 An embedded source route is sometimes encoded in the
3156 "local-part" using "%" as a right-binding routing
3157 operator. For example, in:
3158
3159 user%domain%relay3%relay2@relay1
3160
3161 the "%" convention implies that the mail is to be routed
3162 from "relay1" through "relay2", "relay3", and finally to
3163 "user" at "domain". This is commonly known as the "%-
3164 hack". It is suggested that "%" have lower precedence
3165 than any other routing operator (e.g., "!") hidden in the
3166 local-part; for example, "a!b%c" would be interpreted as
3167 "(a!b)%c".
3168
3169 Only the target host (in this case, "relay1") is permitted
3170 to analyze the local-part "user%domain%relay3%relay2".
3171
3172 5.2.17 Domain Literals: <a href="./rfc822#section-6.2.3">RFC-822 Section&nbsp;6.2.3</a>
3173
3174 A mailer MUST be able to accept and parse an Internet domain
3175 literal whose content ("dtext"; see <a href="./rfc822">RFC-822</a>) is a dotted-
3176 decimal host address. This satisfies the requirement of
3177 <a href="#section-2.1">Section 2.1</a> for the case of mail.
3178
3179 An SMTP MUST accept and recognize a domain literal for any of
3180 its own IP addresses.
3181
3182
3183
3184
3185
3186
3187
3188 <span class="grey">Internet Engineering Task Force [Page 57]</span></pre>
3189 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-58" ></span>
3190 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3191
3192
3193 5.2.18 Common Address Formatting Errors: <a href="./rfc822#section-6.1">RFC-822 Section&nbsp;6.1</a>
3194
3195 Errors in formatting or parsing 822 addresses are unfortunately
3196 common. This section mentions only the most common errors. A
3197 User Agent MUST accept all valid <a href="./rfc822">RFC-822</a> address formats, and
3198 MUST NOT generate illegal address syntax.
3199
3200 o A common error is to leave out the semicolon after a group
3201 identifier.
3202
3203 o Some systems fail to fully-qualify domain names in
3204 messages they generate. The right-hand side of an "@"
3205 sign in a header address field MUST be a fully-qualified
3206 domain name.
3207
3208 For example, some systems fail to fully-qualify the From:
3209 address; this prevents a "reply" command in the user
3210 interface from automatically constructing a return
3211 address.
3212
3213 DISCUSSION:
3214 Although <a href="./rfc822">RFC-822</a> allows the local use of abbreviated
3215 domain names within a domain, the application of
3216 <a href="./rfc822">RFC-822</a> in Internet mail does not allow this. The
3217 intent is that an Internet host must not send an SMTP
3218 message header containing an abbreviated domain name
3219 in an address field. This allows the address fields
3220 of the header to be passed without alteration across
3221 the Internet, as required in <a href="#section-5.2.6">Section 5.2.6</a>.
3222
3223 o Some systems mis-parse multiple-hop explicit source routes
3224 such as:
3225
3226 @relay1,@relay2,@relay3:user@domain.
3227
3228
3229 o Some systems over-qualify domain names by adding a
3230 trailing dot to some or all domain names in addresses or
3231 message-ids. This violates <a href="./rfc822">RFC-822</a> syntax.
3232
3233
3234 5.2.19 Explicit Source Routes: <a href="./rfc822#section-6.2.7">RFC-822 Section&nbsp;6.2.7</a>
3235
3236 Internet host software SHOULD NOT create an <a href="./rfc822">RFC-822</a> header
3237 containing an address with an explicit source route, but MUST
3238 accept such headers for compatibility with earlier systems.
3239
3240 DISCUSSION:
3241
3242
3243
3244 <span class="grey">Internet Engineering Task Force [Page 58]</span></pre>
3245 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-59" ></span>
3246 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3247
3248
3249 In an understatement, <a href="./rfc822">RFC-822</a> says "The use of explicit
3250 source routing is discouraged". Many hosts implemented
3251 <a href="./rfc822">RFC-822</a> source routes incorrectly, so the syntax cannot be
3252 used unambiguously in practice. Many users feel the
3253 syntax is ugly. Explicit source routes are not needed in
3254 the mail envelope for delivery; see <a href="#section-5.2.6">Section 5.2.6</a>. For
3255 all these reasons, explicit source routes using the <a href="./rfc822">RFC-</a>
3256 <a href="./rfc822">822</a> notations are not to be used in Internet mail headers.
3257
3258 As stated in <a href="#section-5.2.16">Section 5.2.16</a>, it is necessary to allow an
3259 explicit source route to be buried in the local-part of an
3260 address, e.g., using the "%-hack", in order to allow mail
3261 to be gatewayed into another environment in which explicit
3262 source routing is necessary. The vigilant will observe
3263 that there is no way for a User Agent to detect and
3264 prevent the use of such implicit source routing when the
3265 destination is within the Internet. We can only
3266 discourage source routing of any kind within the Internet,
3267 as unnecessary and undesirable.
3268
3269 5.3 SPECIFIC ISSUES
3270
3271 5.3.1 SMTP Queueing Strategies
3272
3273 The common structure of a host SMTP implementation includes
3274 user mailboxes, one or more areas for queueing messages in
3275 transit, and one or more daemon processes for sending and
3276 receiving mail. The exact structure will vary depending on the
3277 needs of the users on the host and the number and size of
3278 mailing lists supported by the host. We describe several
3279 optimizations that have proved helpful, particularly for
3280 mailers supporting high traffic levels.
3281
3282 Any queueing strategy MUST include:
3283
3284 o Timeouts on all activities. See <a href="#section-5.3.2">Section 5.3.2</a>.
3285
3286 o Never sending error messages in response to error
3287 messages.
3288
3289
3290 5.3.1.1 Sending Strategy
3291
3292 The general model of a sender-SMTP is one or more processes
3293 that periodically attempt to transmit outgoing mail. In a
3294 typical system, the program that composes a message has some
3295 method for requesting immediate attention for a new piece of
3296 outgoing mail, while mail that cannot be transmitted
3297
3298
3299
3300 <span class="grey">Internet Engineering Task Force [Page 59]</span></pre>
3301 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-60" ></span>
3302 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3303
3304
3305 immediately MUST be queued and periodically retried by the
3306 sender. A mail queue entry will include not only the
3307 message itself but also the envelope information.
3308
3309 The sender MUST delay retrying a particular destination
3310 after one attempt has failed. In general, the retry
3311 interval SHOULD be at least 30 minutes; however, more
3312 sophisticated and variable strategies will be beneficial
3313 when the sender-SMTP can determine the reason for non-
3314 delivery.
3315
3316 Retries continue until the message is transmitted or the
3317 sender gives up; the give-up time generally needs to be at
3318 least 4-5 days. The parameters to the retry algorithm MUST
3319 be configurable.
3320
3321 A sender SHOULD keep a list of hosts it cannot reach and
3322 corresponding timeouts, rather than just retrying queued
3323 mail items.
3324
3325 DISCUSSION:
3326 Experience suggests that failures are typically
3327 transient (the target system has crashed), favoring a
3328 policy of two connection attempts in the first hour the
3329 message is in the queue, and then backing off to once
3330 every two or three hours.
3331
3332 The sender-SMTP can shorten the queueing delay by
3333 cooperation with the receiver-SMTP. In particular, if
3334 mail is received from a particular address, it is good
3335 evidence that any mail queued for that host can now be
3336 sent.
3337
3338 The strategy may be further modified as a result of
3339 multiple addresses per host (see <a href="#section-5.3.4">Section 5.3.4</a>), to
3340 optimize delivery time vs. resource usage.
3341
3342 A sender-SMTP may have a large queue of messages for
3343 each unavailable destination host, and if it retried
3344 all these messages in every retry cycle, there would be
3345 excessive Internet overhead and the daemon would be
3346 blocked for a long period. Note that an SMTP can
3347 generally determine that a delivery attempt has failed
3348 only after a timeout of a minute or more; a one minute
3349 timeout per connection will result in a very large
3350 delay if it is repeated for dozens or even hundreds of
3351 queued messages.
3352
3353
3354
3355
3356 <span class="grey">Internet Engineering Task Force [Page 60]</span></pre>
3357 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-61" ></span>
3358 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3359
3360
3361 When the same message is to be delivered to several users on
3362 the same host, only one copy of the message SHOULD be
3363 transmitted. That is, the sender-SMTP should use the
3364 command sequence: RCPT, RCPT,... RCPT, DATA instead of the
3365 sequence: RCPT, DATA, RCPT, DATA,... RCPT, DATA.
3366 Implementation of this efficiency feature is strongly urged.
3367
3368 Similarly, the sender-SMTP MAY support multiple concurrent
3369 outgoing mail transactions to achieve timely delivery.
3370 However, some limit SHOULD be imposed to protect the host
3371 from devoting all its resources to mail.
3372
3373 The use of the different addresses of a multihomed host is
3374 discussed below.
3375
3376 5.3.1.2 Receiving strategy
3377
3378 The receiver-SMTP SHOULD attempt to keep a pending listen on
3379 the SMTP port at all times. This will require the support
3380 of multiple incoming TCP connections for SMTP. Some limit
3381 MAY be imposed.
3382
3383 IMPLEMENTATION:
3384 When the receiver-SMTP receives mail from a particular
3385 host address, it could notify the sender-SMTP to retry
3386 any mail pending for that host address.
3387
3388 5.3.2 Timeouts in SMTP
3389
3390 There are two approaches to timeouts in the sender-SMTP: (a)
3391 limit the time for each SMTP command separately, or (b) limit
3392 the time for the entire SMTP dialogue for a single mail
3393 message. A sender-SMTP SHOULD use option (a), per-command
3394 timeouts. Timeouts SHOULD be easily reconfigurable, preferably
3395 without recompiling the SMTP code.
3396
3397 DISCUSSION:
3398 Timeouts are an essential feature of an SMTP
3399 implementation. If the timeouts are too long (or worse,
3400 there are no timeouts), Internet communication failures or
3401 software bugs in receiver-SMTP programs can tie up SMTP
3402 processes indefinitely. If the timeouts are too short,
3403 resources will be wasted with attempts that time out part
3404 way through message delivery.
3405
3406 If option (b) is used, the timeout has to be very large,
3407 e.g., an hour, to allow time to expand very large mailing
3408 lists. The timeout may also need to increase linearly
3409
3410
3411
3412 <span class="grey">Internet Engineering Task Force [Page 61]</span></pre>
3413 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-62" ></span>
3414 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3415
3416
3417 with the size of the message, to account for the time to
3418 transmit a very large message. A large fixed timeout
3419 leads to two problems: a failure can still tie up the
3420 sender for a very long time, and very large messages may
3421 still spuriously time out (which is a wasteful failure!).
3422
3423 Using the recommended option (a), a timer is set for each
3424 SMTP command and for each buffer of the data transfer.
3425 The latter means that the overall timeout is inherently
3426 proportional to the size of the message.
3427
3428 Based on extensive experience with busy mail-relay hosts, the
3429 minimum per-command timeout values SHOULD be as follows:
3430
3431 o Initial 220 Message: 5 minutes
3432
3433 A Sender-SMTP process needs to distinguish between a
3434 failed TCP connection and a delay in receiving the initial
3435 220 greeting message. Many receiver-SMTPs will accept a
3436 TCP connection but delay delivery of the 220 message until
3437 their system load will permit more mail to be processed.
3438
3439 o MAIL Command: 5 minutes
3440
3441
3442 o RCPT Command: 5 minutes
3443
3444 A longer timeout would be required if processing of
3445 mailing lists and aliases were not deferred until after
3446 the message was accepted.
3447
3448 o DATA Initiation: 2 minutes
3449
3450 This is while awaiting the "354 Start Input" reply to a
3451 DATA command.
3452
3453 o Data Block: 3 minutes
3454
3455 This is while awaiting the completion of each TCP SEND
3456 call transmitting a chunk of data.
3457
3458 o DATA Termination: 10 minutes.
3459
3460 This is while awaiting the "250 OK" reply. When the
3461 receiver gets the final period terminating the message
3462 data, it typically performs processing to deliver the
3463 message to a user mailbox. A spurious timeout at this
3464 point would be very wasteful, since the message has been
3465
3466
3467
3468 <span class="grey">Internet Engineering Task Force [Page 62]</span></pre>
3469 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-63" ></span>
3470 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3471
3472
3473 successfully sent.
3474
3475 A receiver-SMTP SHOULD have a timeout of at least 5 minutes
3476 while it is awaiting the next command from the sender.
3477
3478 5.3.3 Reliable Mail Receipt
3479
3480 When the receiver-SMTP accepts a piece of mail (by sending a
3481 "250 OK" message in response to DATA), it is accepting
3482 responsibility for delivering or relaying the message. It must
3483 take this responsibility seriously, i.e., it MUST NOT lose the
3484 message for frivolous reasons, e.g., because the host later
3485 crashes or because of a predictable resource shortage.
3486
3487 If there is a delivery failure after acceptance of a message,
3488 the receiver-SMTP MUST formulate and mail a notification
3489 message. This notification MUST be sent using a null ("&lt;&gt;")
3490 reverse path in the envelope; see <a href="./rfc821#section-3.6">Section&nbsp;3.6 of RFC-821</a>. The
3491 recipient of this notification SHOULD be the address from the
3492 envelope return path (or the Return-Path: line). However, if
3493 this address is null ("&lt;&gt;"), the receiver-SMTP MUST NOT send a
3494 notification. If the address is an explicit source route, it
3495 SHOULD be stripped down to its final hop.
3496
3497 DISCUSSION:
3498 For example, suppose that an error notification must be
3499 sent for a message that arrived with:
3500 "MAIL FROM:&lt;@a,@b:user@d&gt;". The notification message
3501 should be sent to: "RCPT TO:&lt;user@d&gt;".
3502
3503 Some delivery failures after the message is accepted by
3504 SMTP will be unavoidable. For example, it may be
3505 impossible for the receiver-SMTP to validate all the
3506 delivery addresses in RCPT command(s) due to a "soft"
3507 domain system error or because the target is a mailing
3508 list (see earlier discussion of RCPT).
3509
3510 To avoid receiving duplicate messages as the result of
3511 timeouts, a receiver-SMTP MUST seek to minimize the time
3512 required to respond to the final "." that ends a message
3513 transfer. See <a href="./rfc1047">RFC-1047</a> [SMTP:4] for a discussion of this
3514 problem.
3515
3516 5.3.4 Reliable Mail Transmission
3517
3518 To transmit a message, a sender-SMTP determines the IP address
3519 of the target host from the destination address in the
3520 envelope. Specifically, it maps the string to the right of the
3521
3522
3523
3524 <span class="grey">Internet Engineering Task Force [Page 63]</span></pre>
3525 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-64" ></span>
3526 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3527
3528
3529 "@" sign into an IP address. This mapping or the transfer
3530 itself may fail with a soft error, in which case the sender-
3531 SMTP will requeue the outgoing mail for a later retry, as
3532 required in <a href="#section-5.3.1.1">Section 5.3.1.1</a>.
3533
3534 When it succeeds, the mapping can result in a list of
3535 alternative delivery addresses rather than a single address,
3536 because of (a) multiple MX records, (b) multihoming, or both.
3537 To provide reliable mail transmission, the sender-SMTP MUST be
3538 able to try (and retry) each of the addresses in this list in
3539 order, until a delivery attempt succeeds. However, there MAY
3540 also be a configurable limit on the number of alternate
3541 addresses that can be tried. In any case, a host SHOULD try at
3542 least two addresses.
3543
3544 The following information is to be used to rank the host
3545 addresses:
3546
3547 (1) Multiple MX Records -- these contain a preference
3548 indication that should be used in sorting. If there are
3549 multiple destinations with the same preference and there
3550 is no clear reason to favor one (e.g., by address
3551 preference), then the sender-SMTP SHOULD pick one at
3552 random to spread the load across multiple mail exchanges
3553 for a specific organization; note that this is a
3554 refinement of the procedure in [DNS:3].
3555
3556 (2) Multihomed host -- The destination host (perhaps taken
3557 from the preferred MX record) may be multihomed, in which
3558 case the domain name resolver will return a list of
3559 alternative IP addresses. It is the responsibility of the
3560 domain name resolver interface (see <a href="#section-6.1.3.4">Section 6.1.3.4</a> below)
3561 to have ordered this list by decreasing preference, and
3562 SMTP MUST try them in the order presented.
3563
3564 DISCUSSION:
3565 Although the capability to try multiple alternative
3566 addresses is required, there may be circumstances where
3567 specific installations want to limit or disable the use of
3568 alternative addresses. The question of whether a sender
3569 should attempt retries using the different addresses of a
3570 multihomed host has been controversial. The main argument
3571 for using the multiple addresses is that it maximizes the
3572 probability of timely delivery, and indeed sometimes the
3573 probability of any delivery; the counter argument is that
3574 it may result in unnecessary resource use.
3575
3576 Note that resource use is also strongly determined by the
3577
3578
3579
3580 <span class="grey">Internet Engineering Task Force [Page 64]</span></pre>
3581 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-65" ></span>
3582 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3583
3584
3585 sending strategy discussed in <a href="#section-5.3.1">Section 5.3.1</a>.
3586
3587 5.3.5 Domain Name Support
3588
3589 SMTP implementations MUST use the mechanism defined in <a href="#section-6.1">Section</a>
3590 <a href="#section-6.1">6.1</a> for mapping between domain names and IP addresses. This
3591 means that every Internet SMTP MUST include support for the
3592 Internet DNS.
3593
3594 In particular, a sender-SMTP MUST support the MX record scheme
3595 [SMTP:3]. See also <a href="#section-7.4">Section 7.4</a> of [DNS:2] for information on
3596 domain name support for SMTP.
3597
3598 5.3.6 Mailing Lists and Aliases
3599
3600 An SMTP-capable host SHOULD support both the alias and the list
3601 form of address expansion for multiple delivery. When a
3602 message is delivered or forwarded to each address of an
3603 expanded list form, the return address in the envelope
3604 ("MAIL FROM:") MUST be changed to be the address of a person
3605 who administers the list, but the message header MUST be left
3606 unchanged; in particular, the "From" field of the message is
3607 unaffected.
3608
3609 DISCUSSION:
3610 An important mail facility is a mechanism for multi-
3611 destination delivery of a single message, by transforming
3612 or "expanding" a pseudo-mailbox address into a list of
3613 destination mailbox addresses. When a message is sent to
3614 such a pseudo-mailbox (sometimes called an "exploder"),
3615 copies are forwarded or redistributed to each mailbox in
3616 the expanded list. We classify such a pseudo-mailbox as
3617 an "alias" or a "list", depending upon the expansion
3618 rules:
3619
3620 (a) Alias
3621
3622 To expand an alias, the recipient mailer simply
3623 replaces the pseudo-mailbox address in the envelope
3624 with each of the expanded addresses in turn; the rest
3625 of the envelope and the message body are left
3626 unchanged. The message is then delivered or
3627 forwarded to each expanded address.
3628
3629 (b) List
3630
3631 A mailing list may be said to operate by
3632 "redistribution" rather than by "forwarding". To
3633
3634
3635
3636 <span class="grey">Internet Engineering Task Force [Page 65]</span></pre>
3637 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-66" ></span>
3638 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3639
3640
3641 expand a list, the recipient mailer replaces the
3642 pseudo-mailbox address in the envelope with each of
3643 the expanded addresses in turn. The return address in
3644 the envelope is changed so that all error messages
3645 generated by the final deliveries will be returned to
3646 a list administrator, not to the message originator,
3647 who generally has no control over the contents of the
3648 list and will typically find error messages annoying.
3649
3650
3651 5.3.7 Mail Gatewaying
3652
3653 Gatewaying mail between different mail environments, i.e.,
3654 different mail formats and protocols, is complex and does not
3655 easily yield to standardization. See for example [SMTP:5a],
3656 [SMTP:5b]. However, some general requirements may be given for
3657 a gateway between the Internet and another mail environment.
3658
3659 (A) Header fields MAY be rewritten when necessary as messages
3660 are gatewayed across mail environment boundaries.
3661
3662 DISCUSSION:
3663 This may involve interpreting the local-part of the
3664 destination address, as suggested in <a href="#section-5.2.16">Section 5.2.16</a>.
3665
3666 The other mail systems gatewayed to the Internet
3667 generally use a subset of <a href="./rfc822">RFC-822</a> headers, but some
3668 of them do not have an equivalent to the SMTP
3669 envelope. Therefore, when a message leaves the
3670 Internet environment, it may be necessary to fold the
3671 SMTP envelope information into the message header. A
3672 possible solution would be to create new header
3673 fields to carry the envelope information (e.g., "X-
3674 SMTP-MAIL:" and "X-SMTP-RCPT:"); however, this would
3675 require changes in mail programs in the foreign
3676 environment.
3677
3678 (B) When forwarding a message into or out of the Internet
3679 environment, a gateway MUST prepend a Received: line, but
3680 it MUST NOT alter in any way a Received: line that is
3681 already in the header.
3682
3683 DISCUSSION:
3684 This requirement is a subset of the general
3685 "Received:" line requirement of <a href="#section-5.2.8">Section 5.2.8</a>; it is
3686 restated here for emphasis.
3687
3688 Received: fields of messages originating from other
3689
3690
3691
3692 <span class="grey">Internet Engineering Task Force [Page 66]</span></pre>
3693 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-67" ></span>
3694 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3695
3696
3697 environments may not conform exactly to <a href="./rfc822">RFC822</a>.
3698 However, the most important use of Received: lines is
3699 for debugging mail faults, and this debugging can be
3700 severely hampered by well-meaning gateways that try
3701 to "fix" a Received: line.
3702
3703 The gateway is strongly encouraged to indicate the
3704 environment and protocol in the "via" clauses of
3705 Received field(s) that it supplies.
3706
3707 (C) From the Internet side, the gateway SHOULD accept all
3708 valid address formats in SMTP commands and in <a href="./rfc822">RFC-822</a>
3709 headers, and all valid <a href="./rfc822">RFC-822</a> messages. Although a
3710 gateway must accept an <a href="./rfc822">RFC-822</a> explicit source route
3711 ("@...:" format) in either the <a href="./rfc822">RFC-822</a> header or in the
3712 envelope, it MAY or may not act on the source route; see
3713 Sections <a href="#section-5.2.6">5.2.6</a> and <a href="#section-5.2.19">5.2.19</a>.
3714
3715 DISCUSSION:
3716 It is often tempting to restrict the range of
3717 addresses accepted at the mail gateway to simplify
3718 the translation into addresses for the remote
3719 environment. This practice is based on the
3720 assumption that mail users have control over the
3721 addresses their mailers send to the mail gateway. In
3722 practice, however, users have little control over the
3723 addresses that are finally sent; their mailers are
3724 free to change addresses into any legal <a href="./rfc822">RFC-822</a>
3725 format.
3726
3727 (D) The gateway MUST ensure that all header fields of a
3728 message that it forwards into the Internet meet the
3729 requirements for Internet mail. In particular, all
3730 addresses in "From:", "To:", "Cc:", etc., fields must be
3731 transformed (if necessary) to satisfy <a href="./rfc822">RFC-822</a> syntax, and
3732 they must be effective and useful for sending replies.
3733
3734
3735 (E) The translation algorithm used to convert mail from the
3736 Internet protocols to another environment's protocol
3737 SHOULD try to ensure that error messages from the foreign
3738 mail environment are delivered to the return path from the
3739 SMTP envelope, not to the sender listed in the "From:"
3740 field of the <a href="./rfc822">RFC-822</a> message.
3741
3742 DISCUSSION:
3743 Internet mail lists usually place the address of the
3744 mail list maintainer in the envelope but leave the
3745
3746
3747
3748 <span class="grey">Internet Engineering Task Force [Page 67]</span></pre>
3749 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-68" ></span>
3750 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3751
3752
3753 original message header intact (with the "From:"
3754 field containing the original sender). This yields
3755 the behavior the average recipient expects: a reply
3756 to the header gets sent to the original sender, not
3757 to a mail list maintainer; however, errors get sent
3758 to the maintainer (who can fix the problem) and not
3759 the sender (who probably cannot).
3760
3761 (F) Similarly, when forwarding a message from another
3762 environment into the Internet, the gateway SHOULD set the
3763 envelope return path in accordance with an error message
3764 return address, if any, supplied by the foreign
3765 environment.
3766
3767
3768 5.3.8 Maximum Message Size
3769
3770 Mailer software MUST be able to send and receive messages of at
3771 least 64K bytes in length (including header), and a much larger
3772 maximum size is highly desirable.
3773
3774 DISCUSSION:
3775 Although SMTP does not define the maximum size of a
3776 message, many systems impose implementation limits.
3777
3778 The current de facto minimum limit in the Internet is 64K
3779 bytes. However, electronic mail is used for a variety of
3780 purposes that create much larger messages. For example,
3781 mail is often used instead of FTP for transmitting ASCII
3782 files, and in particular to transmit entire documents. As
3783 a result, messages can be 1 megabyte or even larger. We
3784 note that the present document together with its lower-
3785 layer companion contains 0.5 megabytes.
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804 <span class="grey">Internet Engineering Task Force [Page 68]</span></pre>
3805 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-69" ></span>
3806 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3807
3808
3809 5.4 SMTP REQUIREMENTS SUMMARY
3810
3811 | | | | |S| |
3812 | | | | |H| |F
3813 | | | | |O|M|o
3814 | | |S| |U|U|o
3815 | | |H| |L|S|t
3816 | |M|O| |D|T|n
3817 | |U|U|M| | |o
3818 | |S|L|A|N|N|t
3819 | |T|D|Y|O|O|t
3820 FEATURE |SECTION | | | |T|T|e
3821 -----------------------------------------------|----------|-|-|-|-|-|--
3822 | | | | | | |
3823 RECEIVER-SMTP: | | | | | | |
3824 Implement VRFY |5.2.3 |x| | | | |
3825 Implement EXPN |5.2.3 | |x| | | |
3826 EXPN, VRFY configurable |5.2.3 | | |x| | |
3827 Implement SEND, SOML, SAML |5.2.4 | | |x| | |
3828 Verify HELO parameter |5.2.5 | | |x| | |
3829 Refuse message with bad HELO |5.2.5 | | | | |x|
3830 Accept explicit src-route syntax in env. |5.2.6 |x| | | | |
3831 Support "postmaster" |5.2.7 |x| | | | |
3832 Process RCPT when received (except lists) |5.2.7 | | |x| | |
3833 Long delay of RCPT responses |5.2.7 | | | | |x|
3834 | | | | | | |
3835 Add Received: line |5.2.8 |x| | | | |
3836 Received: line include domain literal |5.2.8 | |x| | | |
3837 Change previous Received: line |5.2.8 | | | | |x|
3838 Pass Return-Path info (final deliv/gwy) |5.2.8 |x| | | | |
3839 Support empty reverse path |5.2.9 |x| | | | |
3840 Send only official reply codes |5.2.10 | |x| | | |
3841 Send text from <a href="./rfc821">RFC-821</a> when appropriate |5.2.10 | |x| | | |
3842 Delete "." for transparency |5.2.11 |x| | | | |
3843 Accept and recognize self domain literal(s) |5.2.17 |x| | | | |
3844 | | | | | | |
3845 Error message about error message |5.3.1 | | | | |x|
3846 Keep pending listen on SMTP port |5.3.1.2 | |x| | | |
3847 Provide limit on recv concurrency |5.3.1.2 | | |x| | |
3848 Wait at least 5 mins for next sender cmd |5.3.2 | |x| | | |
3849 Avoidable delivery failure after "250 OK" |5.3.3 | | | | |x|
3850 Send error notification msg after accept |5.3.3 |x| | | | |
3851 Send using null return path |5.3.3 |x| | | | |
3852 Send to envelope return path |5.3.3 | |x| | | |
3853 Send to null address |5.3.3 | | | | |x|
3854 Strip off explicit src route |5.3.3 | |x| | | |
3855 Minimize acceptance delay (<a href="./rfc1047">RFC-1047</a>) |5.3.3 |x| | | | |
3856 -----------------------------------------------|----------|-|-|-|-|-|--
3857
3858
3859
3860 <span class="grey">Internet Engineering Task Force [Page 69]</span></pre>
3861 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-70" ></span>
3862 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3863
3864
3865 | | | | | | |
3866 SENDER-SMTP: | | | | | | |
3867 Canonicalized domain names in MAIL, RCPT |5.2.2 |x| | | | |
3868 Implement SEND, SOML, SAML |5.2.4 | | |x| | |
3869 Send valid principal host name in HELO |5.2.5 |x| | | | |
3870 Send explicit source route in RCPT TO: |5.2.6 | | | |x| |
3871 Use only reply code to determine action |5.2.10 |x| | | | |
3872 Use only high digit of reply code when poss. |5.2.10 | |x| | | |
3873 Add "." for transparency |5.2.11 |x| | | | |
3874 | | | | | | |
3875 Retry messages after soft failure |5.3.1.1 |x| | | | |
3876 Delay before retry |5.3.1.1 |x| | | | |
3877 Configurable retry parameters |5.3.1.1 |x| | | | |
3878 Retry once per each queued dest host |5.3.1.1 | |x| | | |
3879 Multiple RCPT's for same DATA |5.3.1.1 | |x| | | |
3880 Support multiple concurrent transactions |5.3.1.1 | | |x| | |
3881 Provide limit on concurrency |5.3.1.1 | |x| | | |
3882 | | | | | | |
3883 Timeouts on all activities |5.3.1 |x| | | | |
3884 Per-command timeouts |5.3.2 | |x| | | |
3885 Timeouts easily reconfigurable |5.3.2 | |x| | | |
3886 Recommended times |5.3.2 | |x| | | |
3887 Try alternate addr's in order |5.3.4 |x| | | | |
3888 Configurable limit on alternate tries |5.3.4 | | |x| | |
3889 Try at least two alternates |5.3.4 | |x| | | |
3890 Load-split across equal MX alternates |5.3.4 | |x| | | |
3891 Use the Domain Name System |5.3.5 |x| | | | |
3892 Support MX records |5.3.5 |x| | | | |
3893 Use WKS records in MX processing |5.2.12 | | | |x| |
3894 -----------------------------------------------|----------|-|-|-|-|-|--
3895 | | | | | | |
3896 MAIL FORWARDING: | | | | | | |
3897 Alter existing header field(s) |5.2.6 | | | |x| |
3898 Implement relay function: 821/<a href="#section-3.6">section 3.6</a> |5.2.6 | | |x| | |
3899 If not, deliver to RHS domain |5.2.6 | |x| | | |
3900 Interpret 'local-part' of addr |5.2.16 | | | | |x|
3901 | | | | | | |
3902 MAILING LISTS AND ALIASES | | | | | | |
3903 Support both |5.3.6 | |x| | | |
3904 Report mail list error to local admin. |5.3.6 |x| | | | |
3905 | | | | | | |
3906 MAIL GATEWAYS: | | | | | | |
3907 Embed foreign mail route in local-part |5.2.16 | | |x| | |
3908 Rewrite header fields when necessary |5.3.7 | | |x| | |
3909 Prepend Received: line |5.3.7 |x| | | | |
3910 Change existing Received: line |5.3.7 | | | | |x|
3911 Accept full <a href="./rfc822">RFC-822</a> on Internet side |5.3.7 | |x| | | |
3912 Act on <a href="./rfc822">RFC-822</a> explicit source route |5.3.7 | | |x| | |
3913
3914
3915
3916 <span class="grey">Internet Engineering Task Force [Page 70]</span></pre>
3917 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-71" ></span>
3918 <span class="grey"><a href="./rfc1123">RFC1123</a> MAIL -- SMTP &amp; <a href="./rfc822">RFC-822</a> October 1989</span>
3919
3920
3921 Send only valid <a href="./rfc822">RFC-822</a> on Internet side |5.3.7 |x| | | | |
3922 Deliver error msgs to envelope addr |5.3.7 | |x| | | |
3923 Set env return path from err return addr |5.3.7 | |x| | | |
3924 | | | | | | |
3925 USER AGENT -- <a href="./rfc822">RFC-822</a> | | | | | | |
3926 Allow user to enter &lt;route&gt; address |5.2.6 | | | |x| |
3927 Support <a href="./rfc1049">RFC-1049</a> Content Type field |5.2.13 | | |x| | |
3928 Use 4-digit years |5.2.14 | |x| | | |
3929 Generate numeric timezones |5.2.14 | |x| | | |
3930 Accept all timezones |5.2.14 |x| | | | |
3931 Use non-num timezones from <a href="./rfc822">RFC-822</a> |5.2.14 |x| | | | |
3932 Omit phrase before route-addr |5.2.15 | | |x| | |
3933 Accept and parse dot.dec. domain literals |5.2.17 |x| | | | |
3934 Accept all <a href="./rfc822">RFC-822</a> address formats |5.2.18 |x| | | | |
3935 Generate invalid <a href="./rfc822">RFC-822</a> address format |5.2.18 | | | | |x|
3936 Fully-qualified domain names in header |5.2.18 |x| | | | |
3937 Create explicit src route in header |5.2.19 | | | |x| |
3938 Accept explicit src route in header |5.2.19 |x| | | | |
3939 | | | | | | |
3940 Send/recv at least 64KB messages |5.3.8 |x| | | | |
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972 <span class="grey">Internet Engineering Task Force [Page 71]</span></pre>
3973 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-72" ></span>
3974 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
3975
3976
3977 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. SUPPORT SERVICES</span>
3978
3979 6.1 DOMAIN NAME TRANSLATION
3980
3981 6.1.1 INTRODUCTION
3982
3983 Every host MUST implement a resolver for the Domain Name System
3984 (DNS), and it MUST implement a mechanism using this DNS
3985 resolver to convert host names to IP addresses and vice-versa
3986 [DNS:1, DNS:2].
3987
3988 In addition to the DNS, a host MAY also implement a host name
3989 translation mechanism that searches a local Internet host
3990 table. See <a href="#section-6.1.3.8">Section 6.1.3.8</a> for more information on this
3991 option.
3992
3993 DISCUSSION:
3994 Internet host name translation was originally performed by
3995 searching local copies of a table of all hosts. This
3996 table became too large to update and distribute in a
3997 timely manner and too large to fit into many hosts, so the
3998 DNS was invented.
3999
4000 The DNS creates a distributed database used primarily for
4001 the translation between host names and host addresses.
4002 Implementation of DNS software is required. The DNS
4003 consists of two logically distinct parts: name servers and
4004 resolvers (although implementations often combine these
4005 two logical parts in the interest of efficiency) [DNS:2].
4006
4007 Domain name servers store authoritative data about certain
4008 sections of the database and answer queries about the
4009 data. Domain resolvers query domain name servers for data
4010 on behalf of user processes. Every host therefore needs a
4011 DNS resolver; some host machines will also need to run
4012 domain name servers. Since no name server has complete
4013 information, in general it is necessary to obtain
4014 information from more than one name server to resolve a
4015 query.
4016
4017 6.1.2 PROTOCOL WALK-THROUGH
4018
4019 An implementor must study references [DNS:1] and [DNS:2]
4020 carefully. They provide a thorough description of the theory,
4021 protocol, and implementation of the domain name system, and
4022 reflect several years of experience.
4023
4024
4025
4026
4027
4028 <span class="grey">Internet Engineering Task Force [Page 72]</span></pre>
4029 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-73" ></span>
4030 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4031
4032
4033 6.1.2.1 Resource Records with Zero TTL: <a href="./rfc1035#section-3.2.1">RFC-1035 Section&nbsp;3.2.1</a>
4034
4035 All DNS name servers and resolvers MUST properly handle RRs
4036 with a zero TTL: return the RR to the client but do not
4037 cache it.
4038
4039 DISCUSSION:
4040 Zero TTL values are interpreted to mean that the RR can
4041 only be used for the transaction in progress, and
4042 should not be cached; they are useful for extremely
4043 volatile data.
4044
4045 6.1.2.2 QCLASS Values: <a href="./rfc1035#section-3.2.5">RFC-1035 Section&nbsp;3.2.5</a>
4046
4047 A query with "QCLASS=*" SHOULD NOT be used unless the
4048 requestor is seeking data from more than one class. In
4049 particular, if the requestor is only interested in Internet
4050 data types, QCLASS=IN MUST be used.
4051
4052 6.1.2.3 Unused Fields: <a href="./rfc1035#section-4.1.1">RFC-1035 Section&nbsp;4.1.1</a>
4053
4054 Unused fields in a query or response message MUST be zero.
4055
4056 6.1.2.4 Compression: <a href="./rfc1035#section-4.1.4">RFC-1035 Section&nbsp;4.1.4</a>
4057
4058 Name servers MUST use compression in responses.
4059
4060 DISCUSSION:
4061 Compression is essential to avoid overflowing UDP
4062 datagrams; see <a href="#section-6.1.3.2">Section 6.1.3.2</a>.
4063
4064 6.1.2.5 Misusing Configuration Info: <a href="./rfc1035#section-6.1.2">RFC-1035 Section&nbsp;6.1.2</a>
4065
4066 Recursive name servers and full-service resolvers generally
4067 have some configuration information containing hints about
4068 the location of root or local name servers. An
4069 implementation MUST NOT include any of these hints in a
4070 response.
4071
4072 DISCUSSION:
4073 Many implementors have found it convenient to store
4074 these hints as if they were cached data, but some
4075 neglected to ensure that this "cached data" was not
4076 included in responses. This has caused serious
4077 problems in the Internet when the hints were obsolete
4078 or incorrect.
4079
4080
4081
4082
4083
4084 <span class="grey">Internet Engineering Task Force [Page 73]</span></pre>
4085 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-74" ></span>
4086 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4087
4088
4089 6.1.3 SPECIFIC ISSUES
4090
4091 6.1.3.1 Resolver Implementation
4092
4093 A name resolver SHOULD be able to multiplex concurrent
4094 requests if the host supports concurrent processes.
4095
4096 In implementing a DNS resolver, one of two different models
4097 MAY optionally be chosen: a full-service resolver, or a stub
4098 resolver.
4099
4100
4101 (A) Full-Service Resolver
4102
4103 A full-service resolver is a complete implementation of
4104 the resolver service, and is capable of dealing with
4105 communication failures, failure of individual name
4106 servers, location of the proper name server for a given
4107 name, etc. It must satisfy the following requirements:
4108
4109 o The resolver MUST implement a local caching
4110 function to avoid repeated remote access for
4111 identical requests, and MUST time out information
4112 in the cache.
4113
4114 o The resolver SHOULD be configurable with start-up
4115 information pointing to multiple root name servers
4116 and multiple name servers for the local domain.
4117 This insures that the resolver will be able to
4118 access the whole name space in normal cases, and
4119 will be able to access local domain information
4120 should the local network become disconnected from
4121 the rest of the Internet.
4122
4123
4124 (B) Stub Resolver
4125
4126 A "stub resolver" relies on the services of a recursive
4127 name server on the connected network or a "nearby"
4128 network. This scheme allows the host to pass on the
4129 burden of the resolver function to a name server on
4130 another host. This model is often essential for less
4131 capable hosts, such as PCs, and is also recommended
4132 when the host is one of several workstations on a local
4133 network, because it allows all of the workstations to
4134 share the cache of the recursive name server and hence
4135 reduce the number of domain requests exported by the
4136 local network.
4137
4138
4139
4140 <span class="grey">Internet Engineering Task Force [Page 74]</span></pre>
4141 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-75" ></span>
4142 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4143
4144
4145 At a minimum, the stub resolver MUST be capable of
4146 directing its requests to redundant recursive name
4147 servers. Note that recursive name servers are allowed
4148 to restrict the sources of requests that they will
4149 honor, so the host administrator must verify that the
4150 service will be provided. Stub resolvers MAY implement
4151 caching if they choose, but if so, MUST timeout cached
4152 information.
4153
4154
4155 6.1.3.2 Transport Protocols
4156
4157 DNS resolvers and recursive servers MUST support UDP, and
4158 SHOULD support TCP, for sending (non-zone-transfer) queries.
4159 Specifically, a DNS resolver or server that is sending a
4160 non-zone-transfer query MUST send a UDP query first. If the
4161 Answer section of the response is truncated and if the
4162 requester supports TCP, it SHOULD try the query again using
4163 TCP.
4164
4165 DNS servers MUST be able to service UDP queries and SHOULD
4166 be able to service TCP queries. A name server MAY limit the
4167 resources it devotes to TCP queries, but it SHOULD NOT
4168 refuse to service a TCP query just because it would have
4169 succeeded with UDP.
4170
4171 Truncated responses MUST NOT be saved (cached) and later
4172 used in such a way that the fact that they are truncated is
4173 lost.
4174
4175 DISCUSSION:
4176 UDP is preferred over TCP for queries because UDP
4177 queries have much lower overhead, both in packet count
4178 and in connection state. The use of UDP is essential
4179 for heavily-loaded servers, especially the root
4180 servers. UDP also offers additional robustness, since
4181 a resolver can attempt several UDP queries to different
4182 servers for the cost of a single TCP query.
4183
4184 It is possible for a DNS response to be truncated,
4185 although this is a very rare occurrence in the present
4186 Internet DNS. Practically speaking, truncation cannot
4187 be predicted, since it is data-dependent. The
4188 dependencies include the number of RRs in the answer,
4189 the size of each RR, and the savings in space realized
4190 by the name compression algorithm. As a rule of thumb,
4191 truncation in NS and MX lists should not occur for
4192 answers containing 15 or fewer RRs.
4193
4194
4195
4196 <span class="grey">Internet Engineering Task Force [Page 75]</span></pre>
4197 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-76" ></span>
4198 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4199
4200
4201 Whether it is possible to use a truncated answer
4202 depends on the application. A mailer must not use a
4203 truncated MX response, since this could lead to mail
4204 loops.
4205
4206 Responsible practices can make UDP suffice in the vast
4207 majority of cases. Name servers must use compression
4208 in responses. Resolvers must differentiate truncation
4209 of the Additional section of a response (which only
4210 loses extra information) from truncation of the Answer
4211 section (which for MX records renders the response
4212 unusable by mailers). Database administrators should
4213 list only a reasonable number of primary names in lists
4214 of name servers, MX alternatives, etc.
4215
4216 However, it is also clear that some new DNS record
4217 types defined in the future will contain information
4218 exceeding the 512 byte limit that applies to UDP, and
4219 hence will require TCP. Thus, resolvers and name
4220 servers should implement TCP services as a backup to
4221 UDP today, with the knowledge that they will require
4222 the TCP service in the future.
4223
4224 By private agreement, name servers and resolvers MAY arrange
4225 to use TCP for all traffic between themselves. TCP MUST be
4226 used for zone transfers.
4227
4228 A DNS server MUST have sufficient internal concurrency that
4229 it can continue to process UDP queries while awaiting a
4230 response or performing a zone transfer on an open TCP
4231 connection [DNS:2].
4232
4233 A server MAY support a UDP query that is delivered using an
4234 IP broadcast or multicast address. However, the Recursion
4235 Desired bit MUST NOT be set in a query that is multicast,
4236 and MUST be ignored by name servers receiving queries via a
4237 broadcast or multicast address. A host that sends broadcast
4238 or multicast DNS queries SHOULD send them only as occasional
4239 probes, caching the IP address(es) it obtains from the
4240 response(s) so it can normally send unicast queries.
4241
4242 DISCUSSION:
4243 Broadcast or (especially) IP multicast can provide a
4244 way to locate nearby name servers without knowing their
4245 IP addresses in advance. However, general broadcasting
4246 of recursive queries can result in excessive and
4247 unnecessary load on both network and servers.
4248
4249
4250
4251
4252 <span class="grey">Internet Engineering Task Force [Page 76]</span></pre>
4253 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-77" ></span>
4254 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4255
4256
4257 6.1.3.3 Efficient Resource Usage
4258
4259 The following requirements on servers and resolvers are very
4260 important to the health of the Internet as a whole,
4261 particularly when DNS services are invoked repeatedly by
4262 higher level automatic servers, such as mailers.
4263
4264 (1) The resolver MUST implement retransmission controls to
4265 insure that it does not waste communication bandwidth,
4266 and MUST impose finite bounds on the resources consumed
4267 to respond to a single request. See [DNS:2] pages 43-
4268 44 for specific recommendations.
4269
4270 (2) After a query has been retransmitted several times
4271 without a response, an implementation MUST give up and
4272 return a soft error to the application.
4273
4274 (3) All DNS name servers and resolvers SHOULD cache
4275 temporary failures, with a timeout period of the order
4276 of minutes.
4277
4278 DISCUSSION:
4279 This will prevent applications that immediately
4280 retry soft failures (in violation of <a href="#section-2.2">Section 2.2</a>
4281 of this document) from generating excessive DNS
4282 traffic.
4283
4284 (4) All DNS name servers and resolvers SHOULD cache
4285 negative responses that indicate the specified name, or
4286 data of the specified type, does not exist, as
4287 described in [DNS:2].
4288
4289 (5) When a DNS server or resolver retries a UDP query, the
4290 retry interval SHOULD be constrained by an exponential
4291 backoff algorithm, and SHOULD also have upper and lower
4292 bounds.
4293
4294 IMPLEMENTATION:
4295 A measured RTT and variance (if available) should
4296 be used to calculate an initial retransmission
4297 interval. If this information is not available, a
4298 default of no less than 5 seconds should be used.
4299 Implementations may limit the retransmission
4300 interval, but this limit must exceed twice the
4301 Internet maximum segment lifetime plus service
4302 delay at the name server.
4303
4304 (6) When a resolver or server receives a Source Quench for
4305
4306
4307
4308 <span class="grey">Internet Engineering Task Force [Page 77]</span></pre>
4309 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-78" ></span>
4310 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4311
4312
4313 a query it has issued, it SHOULD take steps to reduce
4314 the rate of querying that server in the near future. A
4315 server MAY ignore a Source Quench that it receives as
4316 the result of sending a response datagram.
4317
4318 IMPLEMENTATION:
4319 One recommended action to reduce the rate is to
4320 send the next query attempt to an alternate
4321 server, if there is one available. Another is to
4322 backoff the retry interval for the same server.
4323
4324
4325 6.1.3.4 Multihomed Hosts
4326
4327 When the host name-to-address function encounters a host
4328 with multiple addresses, it SHOULD rank or sort the
4329 addresses using knowledge of the immediately connected
4330 network number(s) and any other applicable performance or
4331 history information.
4332
4333 DISCUSSION:
4334 The different addresses of a multihomed host generally
4335 imply different Internet paths, and some paths may be
4336 preferable to others in performance, reliability, or
4337 administrative restrictions. There is no general way
4338 for the domain system to determine the best path. A
4339 recommended approach is to base this decision on local
4340 configuration information set by the system
4341 administrator.
4342
4343 IMPLEMENTATION:
4344 The following scheme has been used successfully:
4345
4346 (a) Incorporate into the host configuration data a
4347 Network-Preference List, that is simply a list of
4348 networks in preferred order. This list may be
4349 empty if there is no preference.
4350
4351 (b) When a host name is mapped into a list of IP
4352 addresses, these addresses should be sorted by
4353 network number, into the same order as the
4354 corresponding networks in the Network-Preference
4355 List. IP addresses whose networks do not appear
4356 in the Network-Preference List should be placed at
4357 the end of the list.
4358
4359
4360
4361
4362
4363
4364 <span class="grey">Internet Engineering Task Force [Page 78]</span></pre>
4365 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-79" ></span>
4366 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4367
4368
4369 6.1.3.5 Extensibility
4370
4371 DNS software MUST support all well-known, class-independent
4372 formats [DNS:2], and SHOULD be written to minimize the
4373 trauma associated with the introduction of new well-known
4374 types and local experimentation with non-standard types.
4375
4376 DISCUSSION:
4377 The data types and classes used by the DNS are
4378 extensible, and thus new types will be added and old
4379 types deleted or redefined. Introduction of new data
4380 types ought to be dependent only upon the rules for
4381 compression of domain names inside DNS messages, and
4382 the translation between printable (i.e., master file)
4383 and internal formats for Resource Records (RRs).
4384
4385 Compression relies on knowledge of the format of data
4386 inside a particular RR. Hence compression must only be
4387 used for the contents of well-known, class-independent
4388 RRs, and must never be used for class-specific RRs or
4389 RR types that are not well-known. The owner name of an
4390 RR is always eligible for compression.
4391
4392 A name server may acquire, via zone transfer, RRs that
4393 the server doesn't know how to convert to printable
4394 format. A resolver can receive similar information as
4395 the result of queries. For proper operation, this data
4396 must be preserved, and hence the implication is that
4397 DNS software cannot use textual formats for internal
4398 storage.
4399
4400 The DNS defines domain name syntax very generally -- a
4401 string of labels each containing up to 63 8-bit octets,
4402 separated by dots, and with a maximum total of 255
4403 octets. Particular applications of the DNS are
4404 permitted to further constrain the syntax of the domain
4405 names they use, although the DNS deployment has led to
4406 some applications allowing more general names. In
4407 particular, <a href="#section-2.1">Section 2.1</a> of this document liberalizes
4408 slightly the syntax of a legal Internet host name that
4409 was defined in <a href="./rfc952">RFC-952</a> [DNS:4].
4410
4411 6.1.3.6 Status of RR Types
4412
4413 Name servers MUST be able to load all RR types except MD and
4414 MF from configuration files. The MD and MF types are
4415 obsolete and MUST NOT be implemented; in particular, name
4416 servers MUST NOT load these types from configuration files.
4417
4418
4419
4420 <span class="grey">Internet Engineering Task Force [Page 79]</span></pre>
4421 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-80" ></span>
4422 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4423
4424
4425 DISCUSSION:
4426 The RR types MB, MG, MR, NULL, MINFO and RP are
4427 considered experimental, and applications that use the
4428 DNS cannot expect these RR types to be supported by
4429 most domains. Furthermore these types are subject to
4430 redefinition.
4431
4432 The TXT and WKS RR types have not been widely used by
4433 Internet sites; as a result, an application cannot rely
4434 on the the existence of a TXT or WKS RR in most
4435 domains.
4436
4437 6.1.3.7 Robustness
4438
4439 DNS software may need to operate in environments where the
4440 root servers or other servers are unavailable due to network
4441 connectivity or other problems. In this situation, DNS name
4442 servers and resolvers MUST continue to provide service for
4443 the reachable part of the name space, while giving temporary
4444 failures for the rest.
4445
4446 DISCUSSION:
4447 Although the DNS is meant to be used primarily in the
4448 connected Internet, it should be possible to use the
4449 system in networks which are unconnected to the
4450 Internet. Hence implementations must not depend on
4451 access to root servers before providing service for
4452 local names.
4453
4454 6.1.3.8 Local Host Table
4455
4456 DISCUSSION:
4457 A host may use a local host table as a backup or
4458 supplement to the DNS. This raises the question of
4459 which takes precedence, the DNS or the host table; the
4460 most flexible approach would make this a configuration
4461 option.
4462
4463 Typically, the contents of such a supplementary host
4464 table will be determined locally by the site. However,
4465 a publically-available table of Internet hosts is
4466 maintained by the DDN Network Information Center (DDN
4467 NIC), with a format documented in [DNS:4]. This table
4468 can be retrieved from the DDN NIC using a protocol
4469 described in [DNS:5]. It must be noted that this table
4470 contains only a small fraction of all Internet hosts.
4471 Hosts using this protocol to retrieve the DDN NIC host
4472 table should use the VERSION command to check if the
4473
4474
4475
4476 <span class="grey">Internet Engineering Task Force [Page 80]</span></pre>
4477 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-81" ></span>
4478 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4479
4480
4481 table has changed before requesting the entire table
4482 with the ALL command. The VERSION identifier should be
4483 treated as an arbitrary string and tested only for
4484 equality; no numerical sequence may be assumed.
4485
4486 The DDN NIC host table includes administrative
4487 information that is not needed for host operation and
4488 is therefore not currently included in the DNS
4489 database; examples include network and gateway entries.
4490 However, much of this additional information will be
4491 added to the DNS in the future. Conversely, the DNS
4492 provides essential services (in particular, MX records)
4493 that are not available from the DDN NIC host table.
4494
4495 6.1.4 DNS USER INTERFACE
4496
4497 6.1.4.1 DNS Administration
4498
4499 This document is concerned with design and implementation
4500 issues in host software, not with administrative or
4501 operational issues. However, administrative issues are of
4502 particular importance in the DNS, since errors in particular
4503 segments of this large distributed database can cause poor
4504 or erroneous performance for many sites. These issues are
4505 discussed in [DNS:6] and [DNS:7].
4506
4507 6.1.4.2 DNS User Interface
4508
4509 Hosts MUST provide an interface to the DNS for all
4510 application programs running on the host. This interface
4511 will typically direct requests to a system process to
4512 perform the resolver function [DNS:1, 6.1:2].
4513
4514 At a minimum, the basic interface MUST support a request for
4515 all information of a specific type and class associated with
4516 a specific name, and it MUST return either all of the
4517 requested information, a hard error code, or a soft error
4518 indication. When there is no error, the basic interface
4519 returns the complete response information without
4520 modification, deletion, or ordering, so that the basic
4521 interface will not need to be changed to accommodate new
4522 data types.
4523
4524 DISCUSSION:
4525 The soft error indication is an essential part of the
4526 interface, since it may not always be possible to
4527 access particular information from the DNS; see <a href="#section-6.1.3.3">Section</a>
4528 <a href="#section-6.1.3.3">6.1.3.3</a>.
4529
4530
4531
4532 <span class="grey">Internet Engineering Task Force [Page 81]</span></pre>
4533 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-82" ></span>
4534 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4535
4536
4537 A host MAY provide other DNS interfaces tailored to
4538 particular functions, transforming the raw domain data into
4539 formats more suited to these functions. In particular, a
4540 host MUST provide a DNS interface to facilitate translation
4541 between host addresses and host names.
4542
4543 6.1.4.3 Interface Abbreviation Facilities
4544
4545 User interfaces MAY provide a method for users to enter
4546 abbreviations for commonly-used names. Although the
4547 definition of such methods is outside of the scope of the
4548 DNS specification, certain rules are necessary to insure
4549 that these methods allow access to the entire DNS name space
4550 and to prevent excessive use of Internet resources.
4551
4552 If an abbreviation method is provided, then:
4553
4554 (a) There MUST be some convention for denoting that a name
4555 is already complete, so that the abbreviation method(s)
4556 are suppressed. A trailing dot is the usual method.
4557
4558 (b) Abbreviation expansion MUST be done exactly once, and
4559 MUST be done in the context in which the name was
4560 entered.
4561
4562
4563 DISCUSSION:
4564 For example, if an abbreviation is used in a mail
4565 program for a destination, the abbreviation should be
4566 expanded into a full domain name and stored in the
4567 queued message with an indication that it is already
4568 complete. Otherwise, the abbreviation might be
4569 expanded with a mail system search list, not the
4570 user's, or a name could grow due to repeated
4571 canonicalizations attempts interacting with wildcards.
4572
4573 The two most common abbreviation methods are:
4574
4575 (1) Interface-level aliases
4576
4577 Interface-level aliases are conceptually implemented as
4578 a list of alias/domain name pairs. The list can be
4579 per-user or per-host, and separate lists can be
4580 associated with different functions, e.g. one list for
4581 host name-to-address translation, and a different list
4582 for mail domains. When the user enters a name, the
4583 interface attempts to match the name to the alias
4584 component of a list entry, and if a matching entry can
4585
4586
4587
4588 <span class="grey">Internet Engineering Task Force [Page 82]</span></pre>
4589 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-83" ></span>
4590 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4591
4592
4593 be found, the name is replaced by the domain name found
4594 in the pair.
4595
4596 Note that interface-level aliases and CNAMEs are
4597 completely separate mechanisms; interface-level aliases
4598 are a local matter while CNAMEs are an Internet-wide
4599 aliasing mechanism which is a required part of any DNS
4600 implementation.
4601
4602 (2) Search Lists
4603
4604 A search list is conceptually implemented as an ordered
4605 list of domain names. When the user enters a name, the
4606 domain names in the search list are used as suffixes to
4607 the user-supplied name, one by one, until a domain name
4608 with the desired associated data is found, or the
4609 search list is exhausted. Search lists often contain
4610 the name of the local host's parent domain or other
4611 ancestor domains. Search lists are often per-user or
4612 per-process.
4613
4614 It SHOULD be possible for an administrator to disable a
4615 DNS search-list facility. Administrative denial may be
4616 warranted in some cases, to prevent abuse of the DNS.
4617
4618 There is danger that a search-list mechanism will
4619 generate excessive queries to the root servers while
4620 testing whether user input is a complete domain name,
4621 lacking a final period to mark it as complete. A
4622 search-list mechanism MUST have one of, and SHOULD have
4623 both of, the following two provisions to prevent this:
4624
4625 (a) The local resolver/name server can implement
4626 caching of negative responses (see <a href="#section-6.1.3.3">Section</a>
4627 <a href="#section-6.1.3.3">6.1.3.3</a>).
4628
4629 (b) The search list expander can require two or more
4630 interior dots in a generated domain name before it
4631 tries using the name in a query to non-local
4632 domain servers, such as the root.
4633
4634 DISCUSSION:
4635 The intent of this requirement is to avoid
4636 excessive delay for the user as the search list is
4637 tested, and more importantly to prevent excessive
4638 traffic to the root and other high-level servers.
4639 For example, if the user supplied a name "X" and
4640 the search list contained the root as a component,
4641
4642
4643
4644 <span class="grey">Internet Engineering Task Force [Page 83]</span></pre>
4645 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-84" ></span>
4646 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4647
4648
4649 a query would have to consult a root server before
4650 the next search list alternative could be tried.
4651 The resulting load seen by the root servers and
4652 gateways near the root would be multiplied by the
4653 number of hosts in the Internet.
4654
4655 The negative caching alternative limits the effect
4656 to the first time a name is used. The interior
4657 dot rule is simpler to implement but can prevent
4658 easy use of some top-level names.
4659
4660
4661 6.1.5 DOMAIN NAME SYSTEM REQUIREMENTS SUMMARY
4662
4663 | | | | |S| |
4664 | | | | |H| |F
4665 | | | | |O|M|o
4666 | | |S| |U|U|o
4667 | | |H| |L|S|t
4668 | |M|O| |D|T|n
4669 | |U|U|M| | |o
4670 | |S|L|A|N|N|t
4671 | |T|D|Y|O|O|t
4672 FEATURE |SECTION | | | |T|T|e
4673 -----------------------------------------------|-----------|-|-|-|-|-|--
4674 GENERAL ISSUES | | | | | | |
4675 | | | | | | |
4676 Implement DNS name-to-address conversion |6.1.1 |x| | | | |
4677 Implement DNS address-to-name conversion |6.1.1 |x| | | | |
4678 Support conversions using host table |6.1.1 | | |x| | |
4679 Properly handle RR with zero TTL |6.1.2.1 |x| | | | |
4680 Use QCLASS=* unnecessarily |6.1.2.2 | |x| | | |
4681 Use QCLASS=IN for Internet class |6.1.2.2 |x| | | | |
4682 Unused fields zero |6.1.2.3 |x| | | | |
4683 Use compression in responses |6.1.2.4 |x| | | | |
4684 | | | | | | |
4685 Include config info in responses |6.1.2.5 | | | | |x|
4686 Support all well-known, class-indep. types |6.1.3.5 |x| | | | |
4687 Easily expand type list |6.1.3.5 | |x| | | |
4688 Load all RR types (except MD and MF) |6.1.3.6 |x| | | | |
4689 Load MD or MF type |6.1.3.6 | | | | |x|
4690 Operate when root servers, etc. unavailable |6.1.3.7 |x| | | | |
4691 -----------------------------------------------|-----------|-|-|-|-|-|--
4692 RESOLVER ISSUES: | | | | | | |
4693 | | | | | | |
4694 Resolver support multiple concurrent requests |6.1.3.1 | |x| | | |
4695 Full-service resolver: |6.1.3.1 | | |x| | |
4696 Local caching |6.1.3.1 |x| | | | |
4697
4698
4699
4700 <span class="grey">Internet Engineering Task Force [Page 84]</span></pre>
4701 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-85" ></span>
4702 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4703
4704
4705 Information in local cache times out |6.1.3.1 |x| | | | |
4706 Configurable with starting info |6.1.3.1 | |x| | | |
4707 Stub resolver: |6.1.3.1 | | |x| | |
4708 Use redundant recursive name servers |6.1.3.1 |x| | | | |
4709 Local caching |6.1.3.1 | | |x| | |
4710 Information in local cache times out |6.1.3.1 |x| | | | |
4711 Support for remote multi-homed hosts: | | | | | | |
4712 Sort multiple addresses by preference list |6.1.3.4 | |x| | | |
4713 | | | | | | |
4714 -----------------------------------------------|-----------|-|-|-|-|-|--
4715 TRANSPORT PROTOCOLS: | | | | | | |
4716 | | | | | | |
4717 Support UDP queries |6.1.3.2 |x| | | | |
4718 Support TCP queries |6.1.3.2 | |x| | | |
4719 Send query using UDP first |6.1.3.2 |x| | | | |1
4720 Try TCP if UDP answers are truncated |6.1.3.2 | |x| | | |
4721 Name server limit TCP query resources |6.1.3.2 | | |x| | |
4722 Punish unnecessary TCP query |6.1.3.2 | | | |x| |
4723 Use truncated data as if it were not |6.1.3.2 | | | | |x|
4724 Private agreement to use only TCP |6.1.3.2 | | |x| | |
4725 Use TCP for zone transfers |6.1.3.2 |x| | | | |
4726 TCP usage not block UDP queries |6.1.3.2 |x| | | | |
4727 Support broadcast or multicast queries |6.1.3.2 | | |x| | |
4728 RD bit set in query |6.1.3.2 | | | | |x|
4729 RD bit ignored by server is b'cast/m'cast |6.1.3.2 |x| | | | |
4730 Send only as occasional probe for addr's |6.1.3.2 | |x| | | |
4731 -----------------------------------------------|-----------|-|-|-|-|-|--
4732 RESOURCE USAGE: | | | | | | |
4733 | | | | | | |
4734 Transmission controls, per [DNS:2] |6.1.3.3 |x| | | | |
4735 Finite bounds per request |6.1.3.3 |x| | | | |
4736 Failure after retries =&gt; soft error |6.1.3.3 |x| | | | |
4737 Cache temporary failures |6.1.3.3 | |x| | | |
4738 Cache negative responses |6.1.3.3 | |x| | | |
4739 Retries use exponential backoff |6.1.3.3 | |x| | | |
4740 Upper, lower bounds |6.1.3.3 | |x| | | |
4741 Client handle Source Quench |6.1.3.3 | |x| | | |
4742 Server ignore Source Quench |6.1.3.3 | | |x| | |
4743 -----------------------------------------------|-----------|-|-|-|-|-|--
4744 USER INTERFACE: | | | | | | |
4745 | | | | | | |
4746 All programs have access to DNS interface |6.1.4.2 |x| | | | |
4747 Able to request all info for given name |6.1.4.2 |x| | | | |
4748 Returns complete info or error |6.1.4.2 |x| | | | |
4749 Special interfaces |6.1.4.2 | | |x| | |
4750 Name&lt;-&gt;Address translation |6.1.4.2 |x| | | | |
4751 | | | | | | |
4752 Abbreviation Facilities: |6.1.4.3 | | |x| | |
4753
4754
4755
4756 <span class="grey">Internet Engineering Task Force [Page 85]</span></pre>
4757 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-86" ></span>
4758 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- DOMAINS October 1989</span>
4759
4760
4761 Convention for complete names |6.1.4.3 |x| | | | |
4762 Conversion exactly once |6.1.4.3 |x| | | | |
4763 Conversion in proper context |6.1.4.3 |x| | | | |
4764 Search list: |6.1.4.3 | | |x| | |
4765 Administrator can disable |6.1.4.3 | |x| | | |
4766 Prevention of excessive root queries |6.1.4.3 |x| | | | |
4767 Both methods |6.1.4.3 | |x| | | |
4768 -----------------------------------------------|-----------|-|-|-|-|-|--
4769 -----------------------------------------------|-----------|-|-|-|-|-|--
4770
4771 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Unless there is private agreement between particular resolver and</span>
4772 <span class="h2"> particular server.</span>
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812 <span class="grey">Internet Engineering Task Force [Page 86]</span></pre>
4813 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-87" ></span>
4814 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- INITIALIZATION October 1989</span>
4815
4816
4817 6.2 HOST INITIALIZATION
4818
4819 6.2.1 INTRODUCTION
4820
4821 This section discusses the initialization of host software
4822 across a connected network, or more generally across an
4823 Internet path. This is necessary for a diskless host, and may
4824 optionally be used for a host with disk drives. For a diskless
4825 host, the initialization process is called "network booting"
4826 and is controlled by a bootstrap program located in a boot ROM.
4827
4828 To initialize a diskless host across the network, there are two
4829 distinct phases:
4830
4831 (1) Configure the IP layer.
4832
4833 Diskless machines often have no permanent storage in which
4834 to store network configuration information, so that
4835 sufficient configuration information must be obtained
4836 dynamically to support the loading phase that follows.
4837 This information must include at least the IP addresses of
4838 the host and of the boot server. To support booting
4839 across a gateway, the address mask and a list of default
4840 gateways are also required.
4841
4842 (2) Load the host system code.
4843
4844 During the loading phase, an appropriate file transfer
4845 protocol is used to copy the system code across the
4846 network from the boot server.
4847
4848 A host with a disk may perform the first step, dynamic
4849 configuration. This is important for microcomputers, whose
4850 floppy disks allow network configuration information to be
4851 mistakenly duplicated on more than one host. Also,
4852 installation of new hosts is much simpler if they automatically
4853 obtain their configuration information from a central server,
4854 saving administrator time and decreasing the probability of
4855 mistakes.
4856
4857 6.2.2 REQUIREMENTS
4858
4859 6.2.2.1 Dynamic Configuration
4860
4861 A number of protocol provisions have been made for dynamic
4862 configuration.
4863
4864 o ICMP Information Request/Reply messages
4865
4866
4867
4868 <span class="grey">Internet Engineering Task Force [Page 87]</span></pre>
4869 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-88" ></span>
4870 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- INITIALIZATION October 1989</span>
4871
4872
4873 This obsolete message pair was designed to allow a host
4874 to find the number of the network it is on.
4875 Unfortunately, it was useful only if the host already
4876 knew the host number part of its IP address,
4877 information that hosts requiring dynamic configuration
4878 seldom had.
4879
4880 o Reverse Address Resolution Protocol (RARP) [BOOT:4]
4881
4882 RARP is a link-layer protocol for a broadcast medium
4883 that allows a host to find its IP address given its
4884 link layer address. Unfortunately, RARP does not work
4885 across IP gateways and therefore requires a RARP server
4886 on every network. In addition, RARP does not provide
4887 any other configuration information.
4888
4889 o ICMP Address Mask Request/Reply messages
4890
4891 These ICMP messages allow a host to learn the address
4892 mask for a particular network interface.
4893
4894 o BOOTP Protocol [BOOT:2]
4895
4896 This protocol allows a host to determine the IP
4897 addresses of the local host and the boot server, the
4898 name of an appropriate boot file, and optionally the
4899 address mask and list of default gateways. To locate a
4900 BOOTP server, the host broadcasts a BOOTP request using
4901 UDP. Ad hoc gateway extensions have been used to
4902 transmit the BOOTP broadcast through gateways, and in
4903 the future the IP Multicasting facility will provide a
4904 standard mechanism for this purpose.
4905
4906
4907 The suggested approach to dynamic configuration is to use
4908 the BOOTP protocol with the extensions defined in "BOOTP
4909 Vendor Information Extensions" <a href="./rfc1084">RFC-1084</a> [BOOT:3]. <a href="./rfc1084">RFC-1084</a>
4910 defines some important general (not vendor-specific)
4911 extensions. In particular, these extensions allow the
4912 address mask to be supplied in BOOTP; we RECOMMEND that the
4913 address mask be supplied in this manner.
4914
4915 DISCUSSION:
4916 Historically, subnetting was defined long after IP, and
4917 so a separate mechanism (ICMP Address Mask messages)
4918 was designed to supply the address mask to a host.
4919 However, the IP address mask and the corresponding IP
4920 address conceptually form a pair, and for operational
4921
4922
4923
4924 <span class="grey">Internet Engineering Task Force [Page 88]</span></pre>
4925 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-89" ></span>
4926 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- INITIALIZATION October 1989</span>
4927
4928
4929 simplicity they ought to be defined at the same time
4930 and by the same mechanism, whether a configuration file
4931 or a dynamic mechanism like BOOTP.
4932
4933 Note that BOOTP is not sufficiently general to specify
4934 the configurations of all interfaces of a multihomed
4935 host. A multihomed host must either use BOOTP
4936 separately for each interface, or configure one
4937 interface using BOOTP to perform the loading, and
4938 perform the complete initialization from a file later.
4939
4940 Application layer configuration information is expected
4941 to be obtained from files after loading of the system
4942 code.
4943
4944 6.2.2.2 Loading Phase
4945
4946 A suggested approach for the loading phase is to use TFTP
4947 [BOOT:1] between the IP addresses established by BOOTP.
4948
4949 TFTP to a broadcast address SHOULD NOT be used, for reasons
4950 explained in <a href="#section-4.2.3.4">Section 4.2.3.4</a>.
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980 <span class="grey">Internet Engineering Task Force [Page 89]</span></pre>
4981 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-90" ></span>
4982 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
4983
4984
4985 6.3 REMOTE MANAGEMENT
4986
4987 6.3.1 INTRODUCTION
4988
4989 The Internet community has recently put considerable effort
4990 into the development of network management protocols. The
4991 result has been a two-pronged approach [MGT:1, MGT:6]: the
4992 Simple Network Management Protocol (SNMP) [MGT:4] and the
4993 Common Management Information Protocol over TCP (CMOT) [MGT:5].
4994
4995 In order to be managed using SNMP or CMOT, a host will need to
4996 implement an appropriate management agent. An Internet host
4997 SHOULD include an agent for either SNMP or CMOT.
4998
4999 Both SNMP and CMOT operate on a Management Information Base
5000 (MIB) that defines a collection of management values. By
5001 reading and setting these values, a remote application may
5002 query and change the state of the managed system.
5003
5004 A standard MIB [MGT:3] has been defined for use by both
5005 management protocols, using data types defined by the Structure
5006 of Management Information (SMI) defined in [MGT:2]. Additional
5007 MIB variables can be introduced under the "enterprises" and
5008 "experimental" subtrees of the MIB naming space [MGT:2].
5009
5010 Every protocol module in the host SHOULD implement the relevant
5011 MIB variables. A host SHOULD implement the MIB variables as
5012 defined in the most recent standard MIB, and MAY implement
5013 other MIB variables when appropriate and useful.
5014
5015 6.3.2 PROTOCOL WALK-THROUGH
5016
5017 The MIB is intended to cover both hosts and gateways, although
5018 there may be detailed differences in MIB application to the two
5019 cases. This section contains the appropriate interpretation of
5020 the MIB for hosts. It is likely that later versions of the MIB
5021 will include more entries for host management.
5022
5023 A managed host must implement the following groups of MIB
5024 object definitions: System, Interfaces, Address Translation,
5025 IP, ICMP, TCP, and UDP.
5026
5027 The following specific interpretations apply to hosts:
5028
5029 o ipInHdrErrors
5030
5031 Note that the error "time-to-live exceeded" can occur in a
5032 host only when it is forwarding a source-routed datagram.
5033
5034
5035
5036 <span class="grey">Internet Engineering Task Force [Page 90]</span></pre>
5037 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-91" ></span>
5038 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5039
5040
5041 o ipOutNoRoutes
5042
5043 This object counts datagrams discarded because no route
5044 can be found. This may happen in a host if all the
5045 default gateways in the host's configuration are down.
5046
5047 o ipFragOKs, ipFragFails, ipFragCreates
5048
5049 A host that does not implement intentional fragmentation
5050 (see "Fragmentation" section of [INTRO:1]) MUST return the
5051 value zero for these three objects.
5052
5053 o icmpOutRedirects
5054
5055 For a host, this object MUST always be zero, since hosts
5056 do not send Redirects.
5057
5058 o icmpOutAddrMaskReps
5059
5060 For a host, this object MUST always be zero, unless the
5061 host is an authoritative source of address mask
5062 information.
5063
5064 o ipAddrTable
5065
5066 For a host, the "IP Address Table" object is effectively a
5067 table of logical interfaces.
5068
5069 o ipRoutingTable
5070
5071 For a host, the "IP Routing Table" object is effectively a
5072 combination of the host's Routing Cache and the static
5073 route table described in "Routing Outbound Datagrams"
5074 section of [INTRO:1].
5075
5076 Within each ipRouteEntry, ipRouteMetric1...4 normally will
5077 have no meaning for a host and SHOULD always be -1, while
5078 ipRouteType will normally have the value "remote".
5079
5080 If destinations on the connected network do not appear in
5081 the Route Cache (see "Routing Outbound Datagrams section
5082 of [INTRO:1]), there will be no entries with ipRouteType
5083 of "direct".
5084
5085
5086 DISCUSSION:
5087 The current MIB does not include Type-of-Service in an
5088 ipRouteEntry, but a future revision is expected to make
5089
5090
5091
5092 <span class="grey">Internet Engineering Task Force [Page 91]</span></pre>
5093 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-92" ></span>
5094 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5095
5096
5097 this addition.
5098
5099 We also expect the MIB to be expanded to allow the remote
5100 management of applications (e.g., the ability to partially
5101 reconfigure mail systems). Network service applications
5102 such as mail systems should therefore be written with the
5103 "hooks" for remote management.
5104
5105 6.3.3 MANAGEMENT REQUIREMENTS SUMMARY
5106
5107 | | | | |S| |
5108 | | | | |H| |F
5109 | | | | |O|M|o
5110 | | |S| |U|U|o
5111 | | |H| |L|S|t
5112 | |M|O| |D|T|n
5113 | |U|U|M| | |o
5114 | |S|L|A|N|N|t
5115 | |T|D|Y|O|O|t
5116 FEATURE |SECTION | | | |T|T|e
5117 -----------------------------------------------|-----------|-|-|-|-|-|--
5118 Support SNMP or CMOT agent |6.3.1 | |x| | | |
5119 Implement specified objects in standard MIB |6.3.1 | |x| | | |
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148 <span class="grey">Internet Engineering Task Force [Page 92]</span></pre>
5149 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-93" ></span>
5150 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5151
5152
5153 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. REFERENCES</span>
5154
5155 This section lists the primary references with which every
5156 implementer must be thoroughly familiar. It also lists some
5157 secondary references that are suggested additional reading.
5158
5159 INTRODUCTORY REFERENCES:
5160
5161
5162 [INTRO:1] "Requirements for Internet Hosts -- Communication Layers,"
5163 IETF Host Requirements Working Group, R. Braden, Ed., <a href="./rfc1122">RFC-1122</a>,
5164 October 1989.
5165
5166 [INTRO:2] "DDN Protocol Handbook," NIC-50004, NIC-50005, NIC-50006,
5167 (three volumes), SRI International, December 1985.
5168
5169 [INTRO:3] "Official Internet Protocols," J. Reynolds and J. Postel,
5170 <a href="./rfc1011">RFC-1011</a>, May 1987.
5171
5172 This document is republished periodically with new RFC numbers;
5173 the latest version must be used.
5174
5175 [INTRO:4] "Protocol Document Order Information," O. Jacobsen and J.
5176 Postel, <a href="./rfc980">RFC-980</a>, March 1986.
5177
5178 [INTRO:5] "Assigned Numbers," J. Reynolds and J. Postel, <a href="./rfc1010">RFC-1010</a>,
5179 May 1987.
5180
5181 This document is republished periodically with new RFC numbers;
5182 the latest version must be used.
5183
5184
5185 TELNET REFERENCES:
5186
5187
5188 [TELNET:1] "Telnet Protocol Specification," J. Postel and J.
5189 Reynolds, <a href="./rfc854">RFC-854</a>, May 1983.
5190
5191 [TELNET:2] "Telnet Option Specification," J. Postel and J. Reynolds,
5192 <a href="./rfc855">RFC-855</a>, May 1983.
5193
5194 [TELNET:3] "Telnet Binary Transmission," J. Postel and J. Reynolds,
5195 <a href="./rfc856">RFC-856</a>, May 1983.
5196
5197 [TELNET:4] "Telnet Echo Option," J. Postel and J. Reynolds, <a href="./rfc857">RFC-857</a>,
5198 May 1983.
5199
5200 [TELNET:5] "Telnet Suppress Go Ahead Option," J. Postel and J.
5201
5202
5203
5204 <span class="grey">Internet Engineering Task Force [Page 93]</span></pre>
5205 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-94" ></span>
5206 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5207
5208
5209 Reynolds, <a href="./rfc858">RFC-858</a>, May 1983.
5210
5211 [TELNET:6] "Telnet Status Option," J. Postel and J. Reynolds, <a href="./rfc859">RFC-</a>
5212 <a href="./rfc859">859</a>, May 1983.
5213
5214 [TELNET:7] "Telnet Timing Mark Option," J. Postel and J. Reynolds,
5215 <a href="./rfc860">RFC-860</a>, May 1983.
5216
5217 [TELNET:8] "Telnet Extended Options List," J. Postel and J.
5218 Reynolds, <a href="./rfc861">RFC-861</a>, May 1983.
5219
5220 [TELNET:9] "Telnet End-Of-Record Option," J. Postel, <a href="./rfc855">RFC-855</a>,
5221 December 1983.
5222
5223 [TELNET:10] "Telnet Terminal-Type Option," J. VanBokkelen, <a href="./rfc1091">RFC-1091</a>,
5224 February 1989.
5225
5226 This document supercedes <a href="./rfc930">RFC-930</a>.
5227
5228 [TELNET:11] "Telnet Window Size Option," D. Waitzman, <a href="./rfc1073">RFC-1073</a>,
5229 October 1988.
5230
5231 [TELNET:12] "Telnet Linemode Option," D. Borman, <a href="./rfc1116">RFC-1116</a>, August
5232 1989.
5233
5234 [TELNET:13] "Telnet Terminal Speed Option," C. Hedrick, <a href="./rfc1079">RFC-1079</a>,
5235 December 1988.
5236
5237 [TELNET:14] "Telnet Remote Flow Control Option," C. Hedrick, <a href="./rfc1080">RFC-</a>
5238 <a href="./rfc1080">1080</a>, November 1988.
5239
5240
5241 SECONDARY TELNET REFERENCES:
5242
5243
5244 [TELNET:15] "Telnet Protocol," MIL-STD-1782, U.S. Department of
5245 Defense, May 1984.
5246
5247 This document is intended to describe the same protocol as <a href="./rfc854">RFC-</a>
5248 <a href="./rfc854">854</a>. In case of conflict, <a href="./rfc854">RFC-854</a> takes precedence, and the
5249 present document takes precedence over both.
5250
5251 [TELNET:16] "SUPDUP Protocol," M. Crispin, <a href="./rfc734">RFC-734</a>, October 1977.
5252
5253 [TELNET:17] "Telnet SUPDUP Option," M. Crispin, <a href="./rfc736">RFC-736</a>, October
5254 1977.
5255
5256 [TELNET:18] "Data Entry Terminal Option," J. Day, <a href="./rfc732">RFC-732</a>, June 1977.
5257
5258
5259
5260 <span class="grey">Internet Engineering Task Force [Page 94]</span></pre>
5261 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-95" ></span>
5262 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5263
5264
5265 [TELNET:19] "TELNET Data Entry Terminal option -- DODIIS
5266 Implementation," A. Yasuda and T. Thompson, <a href="./rfc1043">RFC-1043</a>, February
5267 1988.
5268
5269
5270 FTP REFERENCES:
5271
5272
5273 [FTP:1] "File Transfer Protocol," J. Postel and J. Reynolds, <a href="./rfc959">RFC-</a>
5274 <a href="./rfc959">959</a>, October 1985.
5275
5276 [FTP:2] "Document File Format Standards," J. Postel, <a href="./rfc678">RFC-678</a>,
5277 December 1974.
5278
5279 [FTP:3] "File Transfer Protocol," MIL-STD-1780, U.S. Department of
5280 Defense, May 1984.
5281
5282 This document is based on an earlier version of the FTP
5283 specification (<a href="./rfc765">RFC-765</a>) and is obsolete.
5284
5285
5286 TFTP REFERENCES:
5287
5288
5289 [TFTP:1] "The TFTP Protocol Revision 2," K. Sollins, <a href="./rfc783">RFC-783</a>, June
5290 1981.
5291
5292
5293 MAIL REFERENCES:
5294
5295
5296 [SMTP:1] "Simple Mail Transfer Protocol," J. Postel, <a href="./rfc821">RFC-821</a>, August
5297 1982.
5298
5299 [SMTP:2] "Standard For The Format of ARPA Internet Text Messages,"
5300 D. Crocker, <a href="./rfc822">RFC-822</a>, August 1982.
5301
5302 This document obsoleted an earlier specification, <a href="./rfc733">RFC-733</a>.
5303
5304 [SMTP:3] "Mail Routing and the Domain System," C. Partridge, <a href="./rfc974">RFC-</a>
5305 <a href="./rfc974">974</a>, January 1986.
5306
5307 This RFC describes the use of MX records, a mandatory extension
5308 to the mail delivery process.
5309
5310 [SMTP:4] "Duplicate Messages and SMTP," C. Partridge, <a href="./rfc1047">RFC-1047</a>,
5311 February 1988.
5312
5313
5314
5315
5316 <span class="grey">Internet Engineering Task Force [Page 95]</span></pre>
5317 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-96" ></span>
5318 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5319
5320
5321 [SMTP:5a] "Mapping between X.400 and <a href="./rfc822">RFC 822</a>," S. Kille, <a href="./rfc987">RFC-987</a>,
5322 June 1986.
5323
5324 [SMTP:5b] "Addendum to <a href="./rfc987">RFC-987</a>," S. Kille, RFC-???, September 1987.
5325
5326 The two preceding RFC's define a proposed standard for
5327 gatewaying mail between the Internet and the X.400 environments.
5328
5329 [SMTP:6] "Simple Mail Transfer Protocol," MIL-STD-1781, U.S.
5330 Department of Defense, May 1984.
5331
5332 This specification is intended to describe the same protocol as
5333 does <a href="./rfc821">RFC-821</a>. However, MIL-STD-1781 is incomplete; in
5334 particular, it does not include MX records [SMTP:3].
5335
5336 [SMTP:7] "A Content-Type Field for Internet Messages," M. Sirbu,
5337 <a href="./rfc1049">RFC-1049</a>, March 1988.
5338
5339
5340 DOMAIN NAME SYSTEM REFERENCES:
5341
5342
5343 [DNS:1] "Domain Names - Concepts and Facilities," P. Mockapetris,
5344 <a href="./rfc1034">RFC-1034</a>, November 1987.
5345
5346 This document and the following one obsolete <a href="./rfc882">RFC-882</a>, <a href="./rfc883">RFC-883</a>,
5347 and <a href="./rfc973">RFC-973</a>.
5348
5349 [DNS:2] "Domain Names - Implementation and Specification," <a href="./rfc1035">RFC-1035</a>,
5350 P. Mockapetris, November 1987.
5351
5352
5353 [DNS:3] "Mail Routing and the Domain System," C. Partridge, <a href="./rfc974">RFC-974</a>,
5354 January 1986.
5355
5356
5357 [DNS:4] "DoD Internet Host Table Specification," K. Harrenstein,
5358 <a href="./rfc952">RFC-952</a>, M. Stahl, E. Feinler, October 1985.
5359
5360 SECONDARY DNS REFERENCES:
5361
5362
5363 [DNS:5] "Hostname Server," K. Harrenstein, M. Stahl, E. Feinler,
5364 <a href="./rfc953">RFC-953</a>, October 1985.
5365
5366 [DNS:6] "Domain Administrators Guide," M. Stahl, <a href="./rfc1032">RFC-1032</a>, November
5367 1987.
5368
5369
5370
5371
5372 <span class="grey">Internet Engineering Task Force [Page 96]</span></pre>
5373 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-97" ></span>
5374 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5375
5376
5377 [DNS:7] "Domain Administrators Operations Guide," M. Lottor, <a href="./rfc1033">RFC-</a>
5378 <a href="./rfc1033">1033</a>, November 1987.
5379
5380 [DNS:8] "The Domain Name System Handbook," Vol. 4 of Internet
5381 Protocol Handbook, NIC 50007, SRI Network Information Center,
5382 August 1989.
5383
5384
5385 SYSTEM INITIALIZATION REFERENCES:
5386
5387
5388 [BOOT:1] "Bootstrap Loading Using TFTP," R. Finlayson, <a href="./rfc906">RFC-906</a>, June
5389 1984.
5390
5391 [BOOT:2] "Bootstrap Protocol (BOOTP)," W. Croft and J. Gilmore, <a href="./rfc951">RFC-</a>
5392 <a href="./rfc951">951</a>, September 1985.
5393
5394 [BOOT:3] "BOOTP Vendor Information Extensions," J. Reynolds, <a href="./rfc1084">RFC-</a>
5395 <a href="./rfc1084">1084</a>, December 1988.
5396
5397 Note: this RFC revised and obsoleted <a href="./rfc1048">RFC-1048</a>.
5398
5399 [BOOT:4] "A Reverse Address Resolution Protocol," R. Finlayson, T.
5400 Mann, J. Mogul, and M. Theimer, <a href="./rfc903">RFC-903</a>, June 1984.
5401
5402
5403 MANAGEMENT REFERENCES:
5404
5405
5406 [MGT:1] "IAB Recommendations for the Development of Internet Network
5407 Management Standards," V. Cerf, <a href="./rfc1052">RFC-1052</a>, April 1988.
5408
5409 [MGT:2] "Structure and Identification of Management Information for
5410 TCP/IP-based internets," M. Rose and K. McCloghrie, <a href="./rfc1065">RFC-1065</a>,
5411 August 1988.
5412
5413 [MGT:3] "Management Information Base for Network Management of
5414 TCP/IP-based internets," M. Rose and K. McCloghrie, <a href="./rfc1066">RFC-1066</a>,
5415 August 1988.
5416
5417 [MGT:4] "A Simple Network Management Protocol," J. Case, M. Fedor,
5418 M. Schoffstall, and C. Davin, <a href="./rfc1098">RFC-1098</a>, April 1989.
5419
5420 [MGT:5] "The Common Management Information Services and Protocol
5421 over TCP/IP," U. Warrier and L. Besaw, <a href="./rfc1095">RFC-1095</a>, April 1989.
5422
5423 [MGT:6] "Report of the Second Ad Hoc Network Management Review
5424 Group," V. Cerf, <a href="./rfc1109">RFC-1109</a>, August 1989.
5425
5426
5427
5428 <span class="grey">Internet Engineering Task Force [Page 97]</span></pre>
5429 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-98" ></span>
5430 <span class="grey"><a href="./rfc1123">RFC1123</a> SUPPORT SERVICES -- MANAGEMENT October 1989</span>
5431
5432
5433 Security Considerations
5434
5435 There are many security issues in the application and support
5436 programs of host software, but a full discussion is beyond the scope
5437 of this RFC. Security-related issues are mentioned in sections
5438 concerning TFTP (Sections <a href="#section-4.2.1">4.2.1</a>, <a href="#section-4.2.3.4">4.2.3.4</a>, <a href="#section-4.2.3.5">4.2.3.5</a>), the SMTP VRFY and
5439 EXPN commands (<a href="#section-5.2.3">Section 5.2.3</a>), the SMTP HELO command (5.2.5), and the
5440 SMTP DATA command (<a href="#section-5.2.8">Section 5.2.8</a>).
5441
5442 Author's Address
5443
5444 Robert Braden
5445 USC/Information Sciences Institute
5446 4676 Admiralty Way
5447 Marina del Rey, CA 90292-6695
5448
5449 Phone: (213) 822 1511
5450
5451 EMail: [email protected]
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484 Internet Engineering Task Force [Page 98]
5485 </pre>