| 1 |
pike Module |
| 2 |
|
| 3 |
Bogdan-Andrei Iancu |
| 4 |
|
| 5 |
Voice Sistem SRL |
| 6 |
|
| 7 |
Edited by |
| 8 |
|
| 9 |
Bogdan-Andrei Iancu |
| 10 |
|
| 11 |
Copyright © 2003 FhG FOKUS |
| 12 |
__________________________________________________________________ |
| 13 |
|
| 14 |
Table of Contents |
| 15 |
|
| 16 |
1. Admin Guide |
| 17 |
|
| 18 |
1. Overview |
| 19 |
2. Dependencies |
| 20 |
|
| 21 |
2.1. Kamailio Modules |
| 22 |
2.2. External Libraries or Applications |
| 23 |
|
| 24 |
3. Parameters |
| 25 |
|
| 26 |
3.1. sampling_time_unit (integer) |
| 27 |
3.2. reqs_density_per_unit (integer) |
| 28 |
3.3. remove_latency (integer) |
| 29 |
3.4. pike_log_level (integer) |
| 30 |
|
| 31 |
4. Functions |
| 32 |
|
| 33 |
4.1. pike_check_req() |
| 34 |
4.2. pike_check_ip(ipaddr) |
| 35 |
|
| 36 |
5. RPC Commands |
| 37 |
|
| 38 |
5.1. pike.top |
| 39 |
|
| 40 |
2. RPC calls |
| 41 |
|
| 42 |
1. pike.top |
| 43 |
2. pike.list |
| 44 |
|
| 45 |
3. Developer Guide |
| 46 |
|
| 47 |
List of Examples |
| 48 |
|
| 49 |
1.1. Set sampling_time_unit parameter |
| 50 |
1.2. Set reqs_density_per_unit parameter |
| 51 |
1.3. Set remove_latency parameter |
| 52 |
1.4. Set pike_log_level parameter |
| 53 |
1.5. pike_check_req usage |
| 54 |
1.6. pike_check_ip usage |
| 55 |
2.1. Using pike.top |
| 56 |
3.1. Tree of IP addresses |
| 57 |
|
| 58 |
Chapter 1. Admin Guide |
| 59 |
|
| 60 |
Table of Contents |
| 61 |
|
| 62 |
1. Overview |
| 63 |
2. Dependencies |
| 64 |
|
| 65 |
2.1. Kamailio Modules |
| 66 |
2.2. External Libraries or Applications |
| 67 |
|
| 68 |
3. Parameters |
| 69 |
|
| 70 |
3.1. sampling_time_unit (integer) |
| 71 |
3.2. reqs_density_per_unit (integer) |
| 72 |
3.3. remove_latency (integer) |
| 73 |
3.4. pike_log_level (integer) |
| 74 |
|
| 75 |
4. Functions |
| 76 |
|
| 77 |
4.1. pike_check_req() |
| 78 |
4.2. pike_check_ip(ipaddr) |
| 79 |
|
| 80 |
5. RPC Commands |
| 81 |
|
| 82 |
5.1. pike.top |
| 83 |
|
| 84 |
1. Overview |
| 85 |
|
| 86 |
The pike module keeps trace of all (or selected ones) incoming |
| 87 |
request's IP source and blocks the ones that exceed the limit. It works |
| 88 |
simultaneously for IPv4 and IPv6 addresses. |
| 89 |
|
| 90 |
The module does not implement any actions on blocking - it just simply |
| 91 |
reports that there is high traffic from an IP; what to do, is the |
| 92 |
administrator decision (via scripting). |
| 93 |
|
| 94 |
2. Dependencies |
| 95 |
|
| 96 |
2.1. Kamailio Modules |
| 97 |
2.2. External Libraries or Applications |
| 98 |
|
| 99 |
2.1. Kamailio Modules |
| 100 |
|
| 101 |
The following modules must be loaded before this module: |
| 102 |
* No dependencies on other Kamailio modules. |
| 103 |
|
| 104 |
2.2. External Libraries or Applications |
| 105 |
|
| 106 |
The following libraries or applications must be installed before |
| 107 |
running Kamailio with this module loaded: |
| 108 |
* None. |
| 109 |
|
| 110 |
3. Parameters |
| 111 |
|
| 112 |
3.1. sampling_time_unit (integer) |
| 113 |
3.2. reqs_density_per_unit (integer) |
| 114 |
3.3. remove_latency (integer) |
| 115 |
3.4. pike_log_level (integer) |
| 116 |
|
| 117 |
3.1. sampling_time_unit (integer) |
| 118 |
|
| 119 |
Time period in seconds used for sampling (or the sampling accuracy). |
| 120 |
The smaller the better, but slower. If you want to detect peeks, use a |
| 121 |
small one. To limit the access (like total number of requests on a long |
| 122 |
period of time) to a proxy resource (a gateway for example), use a |
| 123 |
bigger value of this parameter. |
| 124 |
|
| 125 |
IMPORTANT: a too small value may lead to performance penalties due to |
| 126 |
timer process overloading. |
| 127 |
|
| 128 |
Default value is “2”. |
| 129 |
|
| 130 |
Example 1.1. Set sampling_time_unit parameter |
| 131 |
... |
| 132 |
modparam("pike", "sampling_time_unit", 10) |
| 133 |
... |
| 134 |
|
| 135 |
3.2. reqs_density_per_unit (integer) |
| 136 |
|
| 137 |
How many requests should be allowed per sampling_time_unit before |
| 138 |
blocking all the incoming request from that IP. Practically, the |
| 139 |
blocking limit is between ( let's have x=reqs_density_per_unit) x and |
| 140 |
3*x for IPv4 addresses and between x and 8*x for IPv6 addresses. |
| 141 |
|
| 142 |
Default value is 30. |
| 143 |
|
| 144 |
Example 1.2. Set reqs_density_per_unit parameter |
| 145 |
... |
| 146 |
modparam("pike", "reqs_density_per_unit", 30) |
| 147 |
... |
| 148 |
|
| 149 |
3.3. remove_latency (integer) |
| 150 |
|
| 151 |
Specifies for how long the IP address will be kept in memory after the |
| 152 |
last request from that IP address. It's a sort of timeout value, in |
| 153 |
seconds. Note that it is not the duration to keep the IP in state |
| 154 |
'blocked'. An IP is unblocked next occurrence of 'sampling_time_unit' |
| 155 |
that does not exceed 'reqs_density_per_unit'. Keeping an IP in memory |
| 156 |
results in faster reaching of blocked state -- see the notes about the |
| 157 |
limits of getting to state 'blocked'. |
| 158 |
|
| 159 |
Default value is 120. |
| 160 |
|
| 161 |
Example 1.3. Set remove_latency parameter |
| 162 |
... |
| 163 |
modparam("pike", "remove_latency", 130) |
| 164 |
... |
| 165 |
|
| 166 |
3.4. pike_log_level (integer) |
| 167 |
|
| 168 |
Syslog log level to be used by module to auto report the blocking (only |
| 169 |
first time) and unblocking of IPs detected as source of floods. |
| 170 |
|
| 171 |
Default value is 1 (L_WARN). |
| 172 |
|
| 173 |
Example 1.4. Set pike_log_level parameter |
| 174 |
... |
| 175 |
modparam("pike", "pike_log_level", -1) |
| 176 |
... |
| 177 |
|
| 178 |
4. Functions |
| 179 |
|
| 180 |
4.1. pike_check_req() |
| 181 |
4.2. pike_check_ip(ipaddr) |
| 182 |
|
| 183 |
4.1. pike_check_req() |
| 184 |
|
| 185 |
Process the source IP of the current request and return false if the IP |
| 186 |
was exceeding the blocking limit. |
| 187 |
|
| 188 |
Return codes: |
| 189 |
* 1 (true) - IP is not to be blocked or internal error occurred. |
| 190 |
|
| 191 |
Warning |
| 192 |
IMPORTANT: in case of internal error, the function returns true to |
| 193 |
avoid reporting the current processed IP as blocked. |
| 194 |
* -1 (false) - IP is source of flooding, previously detected |
| 195 |
* -2 (false) - IP is detected as a new source of flooding - first |
| 196 |
time detection |
| 197 |
|
| 198 |
This function can be used from REQUEST_ROUTE|ONREPLY_ROUTE. |
| 199 |
|
| 200 |
Example 1.5. pike_check_req usage |
| 201 |
... |
| 202 |
if (!pike_check_req()) { exit; }; |
| 203 |
... |
| 204 |
|
| 205 |
4.2. pike_check_ip(ipaddr) |
| 206 |
|
| 207 |
Process the IP address parameter and return false if it was exceeding |
| 208 |
the blocking limit. The return codes are the same from |
| 209 |
pike_check_req(). |
| 210 |
|
| 211 |
The parameter can contain variables. |
| 212 |
|
| 213 |
This function can be used from REQUEST_ROUTE|ONREPLY_ROUTE. |
| 214 |
|
| 215 |
Example 1.6. pike_check_ip usage |
| 216 |
... |
| 217 |
if (!pike_check_ip("1.2.3.4")) { exit; }; |
| 218 |
... |
| 219 |
if (!pike_check_ip("$si")) { exit; }; |
| 220 |
... |
| 221 |
|
| 222 |
5. RPC Commands |
| 223 |
|
| 224 |
5.1. pike.top |
| 225 |
|
| 226 |
5.1. pike.top |
| 227 |
|
| 228 |
Lists nodes in the pike tree. |
| 229 |
|
| 230 |
Name: pike.top |
| 231 |
|
| 232 |
Parameters: filter (optional) - it can be "ALL", "HOT" or "WARM". If |
| 233 |
missing, the "HOT" nodes are listed. |
| 234 |
|
| 235 |
RPC Command Example: |
| 236 |
... |
| 237 |
kamctl rpc pike.top |
| 238 |
... |
| 239 |
|
| 240 |
Chapter 2. RPC calls |
| 241 |
|
| 242 |
Table of Contents |
| 243 |
|
| 244 |
1. pike.top |
| 245 |
2. pike.list |
| 246 |
|
| 247 |
1. pike.top |
| 248 |
|
| 249 |
Pike.top behaves like a 'top' command and shows source IP addresses of |
| 250 |
incoming requestes to pike_check_req() function. |
| 251 |
|
| 252 |
The IP list is sorted by sum of leaf hits (prev and curr) descending |
| 253 |
and in second level by hits. |
| 254 |
|
| 255 |
Some IPs could be marked as HOT depending on theirs request rates. |
| 256 |
|
| 257 |
pike.top command can take one string parameter which specifies what |
| 258 |
kind of nodes you are interested in. Possible values are HOT or ALL. If |
| 259 |
no argument is given, it behaves as HOT was used. |
| 260 |
|
| 261 |
Marking nodes HOT is done on server side, client only presents given |
| 262 |
data and make some postprocessing like sorting. |
| 263 |
|
| 264 |
Output of this command is a simple dump of ip_tree nodes marked as |
| 265 |
ip-leafs. |
| 266 |
|
| 267 |
Example 2.1. Using pike.top |
| 268 |
... |
| 269 |
kamcli rpc pike.top ALL |
| 270 |
... |
| 271 |
|
| 272 |
2. pike.list |
| 273 |
|
| 274 |
Alias for "pike.top" command. |
| 275 |
|
| 276 |
Chapter 3. Developer Guide |
| 277 |
|
| 278 |
One single tree (for both IPv4 and IPv6) is used. Each node contains a |
| 279 |
byte, the IP addresses stretching from root to the leafs. |
| 280 |
|
| 281 |
Example 3.1. Tree of IP addresses |
| 282 |
/ 193 - 175 - 132 - 164 |
| 283 |
tree root / \ - 142 |
| 284 |
\ 195 - 37 - 78 - 163 |
| 285 |
\ - 79 - 134 |
| 286 |
|
| 287 |
To detect the whole address, step by step, from the root to the leafs, |
| 288 |
the nodes corresponding to each byte of the ip address are expanded. In |
| 289 |
order to be expended a node has to be hit for a given number of times |
| 290 |
(possible by different addresses; in the previous example, the node |
| 291 |
“37” was expended by the 195.37.78.163 and 195.37.79.134 hits). |
| 292 |
|
| 293 |
For 193.175.132.164 with x= reqs_density_per_unit: |
| 294 |
* After first req hits -> the “193” node is built. |
| 295 |
* After x more hits, the “175” node is built; the hits of “193” node |
| 296 |
are split between itself and its child--both of them gone have x/2. |
| 297 |
* And so on for node “132” and “164”. |
| 298 |
* Once “164” build the entire address can be found in the tree. “164” |
| 299 |
becomes a leaf. After it will be hit as a leaf for x times, it will |
| 300 |
become “RED” (further request from this address will be blocked). |
| 301 |
|
| 302 |
So, to build and block this address were needed 3*x hits. Now, if reqs |
| 303 |
start coming from 193.175.132.142, the first 3 bytes are already in the |
| 304 |
tree (they are shared with the previous address), so I will need only x |
| 305 |
hits (to build node “142” and to make it “RED”) to make this address |
| 306 |
also to be blocked. This is the reason for the variable number of hits |
| 307 |
necessary to block an IP. |
| 308 |
|
| 309 |
The maximum number of hits to turn an address red are (n is the |
| 310 |
address' number of bytes): |
| 311 |
|
| 312 |
1 (first byte) + x (second byte) + (x / 2) * (n - 2) (for the rest of |
| 313 |
the bytes) + (n - 1) (to turn the node to red). |
| 314 |
|
| 315 |
So, for IPv4 (n = 4) will be 3x and for IPv6 (n = 16) will be 9x. The |
| 316 |
minimum number of hits to turn an address red is x. |