| 1 |
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" |
| 2 |
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
| 3 |
|
| 4 |
<html xmlns="http://www.w3.org/1999/xhtml"> |
| 5 |
<head> |
| 6 |
<meta http-equiv="content-type" content="text/html; charset=iso-8859-1" /> |
| 7 |
<title>Reasons to prefer logback</title> |
| 8 |
<link rel="stylesheet" type="text/css" href="css/common.css" /> |
| 9 |
<link rel="stylesheet" type="text/css" href="css/screen.css" media="screen" /> |
| 10 |
<link rel="stylesheet" type="text/css" href="css/_print.css" media="print" /> |
| 11 |
|
| 12 |
</head> |
| 13 |
<body onload="decorate();"> |
| 14 |
<script type="text/javascript">prefix='';</script> |
| 15 |
<script src="templates/header.js" type="text/javascript"></script> |
| 16 |
<script type="text/javascript" src="js/jquery-min.js"></script> |
| 17 |
<script type="text/javascript" src="js/decorator.js"></script> |
| 18 |
<div id="left"> |
| 19 |
<noscript>Please turn on Javascript to view this menu</noscript> |
| 20 |
<script src="templates/left.js" type="text/javascript"></script> |
| 21 |
</div> |
| 22 |
<div id="right"> |
| 23 |
<script src="templates/right.js" type="text/javascript"></script> |
| 24 |
</div> |
| 25 |
|
| 26 |
<div id="content"> |
| 27 |
|
| 28 |
<h2>Reasons to prefer logback over log4j</h2> |
| 29 |
|
| 30 |
<p>Logback brings a very large number of improvements over log4j, |
| 31 |
big and small. They are too many to enumerate exhaustively. |
| 32 |
Nevertheless, here is a non-exhaustive list of reasons for |
| 33 |
switching to logback from log4j. Keep in mind that logback is |
| 34 |
conceptually very similar to log4j as both projects were founded |
| 35 |
by the same developer. If you are already familiar with log4j, you |
| 36 |
will quickly feel at home using logback. If you like log4j, you |
| 37 |
will probably love logback.</p> |
| 38 |
|
| 39 |
|
| 40 |
<h3 class="doAnchor" name="fasterImpl">Faster implementation</h3> |
| 41 |
|
| 42 |
<p>Based on our previous work on log4j, logback internals have |
| 43 |
been re-written to perform about ten times faster on certain |
| 44 |
critical execution paths. Not only are logback components faster, |
| 45 |
they have a smaller memory footprint as well.</p> |
| 46 |
|
| 47 |
<h3 class="doAnchor" name="tdd">Extensive battery of tests</h3> |
| 48 |
|
| 49 |
<p>Logback comes with a very extensive battery of tests developed |
| 50 |
over the course of several years and untold hours of work. While |
| 51 |
log4j is also tested, logback takes testing to a completely |
| 52 |
different level. In our opinion, this is the single most important |
| 53 |
reason to prefer logback over log4j. You want your logging |
| 54 |
framework to be rock solid and dependable even under adverse |
| 55 |
conditions. |
| 56 |
</p> |
| 57 |
|
| 58 |
<h3 class="doAnchor" name="slf4j">logback-classic speaks SLF4J |
| 59 |
natively</h3> |
| 60 |
|
| 61 |
<p>Since the <code>Logger</code> class in logback-classic |
| 62 |
implements the SLF4J API natively, you incur zero overhead when |
| 63 |
invoking an SLF4J logger with logback-classic as the underlying |
| 64 |
implementation. Moreover, since logback-classic strongly |
| 65 |
encourages the use of SLF4J as its client API, if you need to |
| 66 |
switch to log4j or to j.u.l., you can do so by replacing one jar |
| 67 |
file with another. You will not need to touch your code logging |
| 68 |
via the SLF4J API. This can drastically reduce the work involved |
| 69 |
in switching logging frameworks. |
| 70 |
</p> |
| 71 |
|
| 72 |
<h3 class="doAnchor" name="docs">Extensive documentation</h3> |
| 73 |
|
| 74 |
<p>Logback ships with detailed and constantly updated |
| 75 |
documentation.</p> |
| 76 |
|
| 77 |
<h3 class="doAnchor" name="DSL">Configuration files in XML or |
| 78 |
Groovy</h3> |
| 79 |
|
| 80 |
<p>The traditional way of configuring logback is via an XML |
| 81 |
file. Most of the examples in the documentation use this XML |
| 82 |
syntax. However, as of logback version 0.9.22, <a |
| 83 |
href="manual/groovy.html">configuration files written in |
| 84 |
Groovy</a> are also supported. Compared to XML, Groovy-style |
| 85 |
configuration is more intuitive, consistent and has a shorter |
| 86 |
syntax. |
| 87 |
</p> |
| 88 |
|
| 89 |
<p> |
| 90 |
There is also a <a |
| 91 |
href="http://logback.qos.ch/translator/asGroovy.html">tool to |
| 92 |
automatically migrate your logback.xml files to |
| 93 |
logback.groovy</a>. |
| 94 |
</p> |
| 95 |
|
| 96 |
<h3 class="doAnchor" name="autoScan">Automatic reloading of |
| 97 |
configuration files</h3> |
| 98 |
|
| 99 |
<p>Logback-classic can <a |
| 100 |
href="manual/configuration.html#autoScan">automatically reload its |
| 101 |
configuration file upon modification</a>. The scanning process is |
| 102 |
fast, contention-free, and dynamically scales to millions of |
| 103 |
invocations per second spread over hundreds of threads. It also |
| 104 |
plays well within application servers and more generally within |
| 105 |
the JEE environment as it does <em>not</em> involve the creation |
| 106 |
of a separate thread for scanning. |
| 107 |
</p> |
| 108 |
|
| 109 |
<h3 class="doAnchor" name="grace">Graceful recovery from I/O |
| 110 |
failures</h3> |
| 111 |
|
| 112 |
<p>Logback's <code>FileAppender</code> and all its sub-classes, |
| 113 |
including <code>RollingFileAppender</code>, can gracefully recover |
| 114 |
from I/O failures. Thus, if a file server fails temporarily, you |
| 115 |
no longer need to restart your application just to get logging |
| 116 |
working again. As soon as the file server comes back up, the |
| 117 |
relevant logback appender will transparently and quickly recover |
| 118 |
from the previous error condition. |
| 119 |
</p> |
| 120 |
|
| 121 |
<h3 class="doAnchor" name="maxHistory">Automatic removal of old |
| 122 |
log archives</h3> |
| 123 |
|
| 124 |
<p>By setting the <span class="option">maxHistory</span> property |
| 125 |
of <a |
| 126 |
href="manual/appenders.html#TimeBasedRollingPolicy">TimeBasedRollingPolicy</a> |
| 127 |
or <a |
| 128 |
href="manual/appenders.html#SizeAndTimeBasedFNATP">SizeAndTimeBasedFNATP</a>, |
| 129 |
you can control the maximum number of archived files. If your |
| 130 |
rolling policy calls for monthly rollover and you wish to keep one |
| 131 |
year's worth of logs, simply set the <span |
| 132 |
class="option">maxHistory</span> property to 12. Archived log |
| 133 |
files older than 12 months will be automatically removed. |
| 134 |
</p> |
| 135 |
|
| 136 |
<h3 class="doAnchor" name="compression">Automatic compression of |
| 137 |
archived log files</h3> |
| 138 |
|
| 139 |
<p><a |
| 140 |
href="manual/appenders.html#RollingFileAppender">RollingFileAppender</a> |
| 141 |
can automatically compress archived log files during |
| 142 |
rollover. Compression always occurs asynchronously so that even |
| 143 |
for large log files, your application is not blocked for the |
| 144 |
duration of the compression. |
| 145 |
</p> |
| 146 |
|
| 147 |
<h3 class="doAnchor" name="prudent">Prudent mode</h3> |
| 148 |
|
| 149 |
<p>In <a href="manual/appenders.html#prudent">prudent mode</a>, |
| 150 |
multiple <code>FileAppender</code> instances running on multiple |
| 151 |
JVMs can safely write to the same log file. With certain |
| 152 |
limitations, prudent mode extends to |
| 153 |
<code>RollingFileAppender</code>. |
| 154 |
</p> |
| 155 |
|
| 156 |
<h3 class="doAnchor" name="lilith">Lilith</h3> |
| 157 |
|
| 158 |
<p><a href="http://lilith.huxhorn.de/">Lilith</a> is a logging and |
| 159 |
access event viewer for logback. It is comparable to log4j's |
| 160 |
chainsaw, except that Lilith is designed to handle large amounts of |
| 161 |
logging data without flinching.</p> |
| 162 |
|
| 163 |
<h3 class="doAnchor" name="conditional">Conditional processing of |
| 164 |
configuration files</h3> |
| 165 |
|
| 166 |
<p>Developers often need to juggle between several logback |
| 167 |
configuration files targeting different environments such as |
| 168 |
development, testing and production. These configuration files |
| 169 |
have substantial parts in common, differing only in a few |
| 170 |
places. To avoid duplication, logback supports <a |
| 171 |
href="manual/configuration.html#conditional">conditional |
| 172 |
processing of configuration files</a> with the help of |
| 173 |
<code><if></code>, <code><then></code> and |
| 174 |
<code><else></code> elements so that a single configuration |
| 175 |
file can adequately target several environments. |
| 176 |
</p> |
| 177 |
|
| 178 |
|
| 179 |
<h3 class="doAnchor" name="filters">Filters</h3> |
| 180 |
|
| 181 |
<p>Logback comes with a wide array of <a |
| 182 |
href="manual/filters.html">filtering capabilities</a> going much |
| 183 |
further than what log4j has to offer. For example, let's assume |
| 184 |
that you have a business-critical application deployed on a |
| 185 |
production server. Given the large volume of transactions |
| 186 |
processed, logging level is set to WARN so that only warnings and |
| 187 |
errors are logged. Now imagine that you are confronted with a bug |
| 188 |
that can be reproduced on the production system but remains |
| 189 |
elusive on the test platform due to unspecified differences |
| 190 |
between those two environments (production/testing). |
| 191 |
</p> |
| 192 |
|
| 193 |
<p>With log4j, your only choice is to lower the logging level to |
| 194 |
DEBUG on the production system in an attempt to identify the |
| 195 |
problem. Unfortunately, this will generate large volume of logging |
| 196 |
data, making analysis difficult. More importantly, extensive |
| 197 |
logging can impact the performance of your application on the |
| 198 |
production system.</p> |
| 199 |
|
| 200 |
<p>With logback, you have the option of keeping logging at the |
| 201 |
WARN level for all users except for the one user, say Alice, who |
| 202 |
is responsible for identifying the problem. When Alice is logged |
| 203 |
on, she will be logging at level DEBUG while other users can |
| 204 |
continue to log at the WARN level. This feat can be accomplished |
| 205 |
by adding 4 lines of XML to your configuration file. Search for |
| 206 |
<code>MDCFilter</code> in the <a |
| 207 |
href="manual/filters.html#TurboFilter">relevant section</a> of the |
| 208 |
manual. |
| 209 |
</p> |
| 210 |
|
| 211 |
|
| 212 |
<h3 class="doAnchor" name="sift">SiftingAppender</h3> |
| 213 |
|
| 214 |
<p><a |
| 215 |
href="manual/appenders.html#SiftingAppender">SiftingAppender</a> |
| 216 |
is an amazingly versatile appender. It can be used to separate (or |
| 217 |
sift) logging according to <b>any</b> given runtime attribute. For |
| 218 |
example, <code>SiftingAppender</code> can separate logging events |
| 219 |
according to user sessions, so that the logs generated by each |
| 220 |
user go into distinct log files, one log file per user. |
| 221 |
</p> |
| 222 |
|
| 223 |
<h3 class="doAnchor" name="packagingData">Stack traces with |
| 224 |
packaging data</h3> |
| 225 |
|
| 226 |
<p>When logback prints an exception, the stack trace will include |
| 227 |
packaging data. Here is a sample stack trace generated by the <a |
| 228 |
href="demo.html">logback-demo</a> web-application.</p> |
| 229 |
|
| 230 |
<pre>14:28:48.835 [btpool0-7] INFO c.q.l.demo.prime.PrimeAction - 99 is not a valid value |
| 231 |
java.lang.Exception: 99 is invalid |
| 232 |
at ch.qos.logback.demo.prime.PrimeAction.execute(PrimeAction.java:28) [classes/:na] |
| 233 |
at org.apache.struts.action.RequestProcessor.processActionPerform(RequestProcessor.java:431) [struts-1.2.9.jar:1.2.9] |
| 234 |
at org.apache.struts.action.RequestProcessor.process(RequestProcessor.java:236) [struts-1.2.9.jar:1.2.9] |
| 235 |
at org.apache.struts.action.ActionServlet.doPost(ActionServlet.java:432) [struts-1.2.9.jar:1.2.9] |
| 236 |
at javax.servlet.http.HttpServlet.service(HttpServlet.java:820) [servlet-api-2.5-6.1.12.jar:6.1.12] |
| 237 |
at org.mortbay.jetty.servlet.ServletHolder.handle(ServletHolder.java:502) [jetty-6.1.12.jar:6.1.12] |
| 238 |
at ch.qos.logback.demo.UserServletFilter.doFilter(UserServletFilter.java:44) [classes/:na] |
| 239 |
at org.mortbay.jetty.servlet.ServletHandler$CachedChain.doFilter(ServletHandler.java:1115) [jetty-6.1.12.jar:6.1.12] |
| 240 |
at org.mortbay.jetty.servlet.ServletHandler.handle(ServletHandler.java:361) [jetty-6.1.12.jar:6.1.12] |
| 241 |
at org.mortbay.jetty.webapp.WebAppContext.handle(WebAppContext.java:417) [jetty-6.1.12.jar:6.1.12] |
| 242 |
at org.mortbay.jetty.handler.ContextHandlerCollection.handle(ContextHandlerCollection.java:230) [jetty-6.1.12.jar:6.1.12]</pre> |
| 243 |
|
| 244 |
<p>From the above, you can recognize that the application is using |
| 245 |
Struts version 1.2.9 and was deployed under jetty version |
| 246 |
6.1.12. Thus, stack traces will quickly inform the reader about |
| 247 |
the classes intervening in the exception but also the package and |
| 248 |
package versions they belong to. When your customers send you a |
| 249 |
stack trace, as a developer you will no longer need to ask them |
| 250 |
to send you information about the versions of packages they are |
| 251 |
using. The information will be part of the stack trace. See <a |
| 252 |
href="manual/layouts.html#xThrowable">"%xThrowable" conversion |
| 253 |
word</a> for details. |
| 254 |
</p> |
| 255 |
|
| 256 |
<p>This feature can be quite helpful to the point that some users |
| 257 |
mistakenly consider it a <a |
| 258 |
href="http://www.jetbrains.net/devnet/message/5259058">feature of |
| 259 |
their IDE</a>. |
| 260 |
</p> |
| 261 |
|
| 262 |
<h3 class="doAnchor" name="logbackAccess">Logback-access, |
| 263 |
i.e. HTTP-access logging with brains, is an integral part of |
| 264 |
logback</h3> |
| 265 |
|
| 266 |
<p>Last but not least, the logback-access module, part of the |
| 267 |
logback distribution, integrates with Servlet containers such as |
| 268 |
Jetty or Tomcat to provide rich and powerful HTTP-access log |
| 269 |
functionality. Since logback-access was part of the initial |
| 270 |
design, all the logback-classic features you love are |
| 271 |
available in logback-access as well.</p> |
| 272 |
|
| 273 |
<h3 class="doAnchor" name="inSummary">In summary</h3> |
| 274 |
|
| 275 |
<p>We have listed a number of reasons for preferring logback over |
| 276 |
log4j. Given that logback builds upon on our previous work on |
| 277 |
log4j, simply put, logback is just a better log4j.</p> |
| 278 |
|
| 279 |
<script src="templates/footer.js" type="text/javascript"></script> |
| 280 |
</div> |
| 281 |
|
| 282 |
</body> |
| 283 |
</html> |