[
https://issues.apache.org/jira/browse/LOG4J2-393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13764286#comment-13764286
]
Gary Gregory commented on LOG4J2-393:
-------------------------------------
That is weird, the method is part of the public Java 6 API:
http://docs.oracle.com/javase/6/docs/api/javax/xml/parsers/DocumentBuilderFactory.html#setFeature%28java.lang.String,%20boolean%29
It's also in Java 5 FWIW so it has been around for a long time:
http://docs.oracle.com/javase/1.5.0/docs/api/javax/xml/parsers/DocumentBuilderFactory.html#setFeature%28java.lang.String,%20boolean%29
What is your Java set up like? Which version and vendor? Do have anything on
the endorsed path? Did you override Xerces or any stock JRE classes?
> Low initialization performance when using Log4J with jar packages
> -----------------------------------------------------------------
>
> Key: LOG4J2-393
> URL: https://issues.apache.org/jira/browse/LOG4J2-393
> Project: Log4j 2
> Issue Type: Bug
> Components: Configurators, Core
> Affects Versions: 2.0-beta8
> Reporter: Philipp Hedwig
> Labels: performance
> Fix For: 2.0-beta9
>
>
> We discovered a huge performance difference when initializing Log4J with
> deployed jar packages compared to using it directly with normal class files.
> As it seems,
> _org.apache.logging.log4j.core.config.plugins.ResolverUtil#loadImplementationsInJar_
> is way slower than its counterpart
> _org.apache.logging.log4j.core.config.plugins.ResolverUtil#loadImplementationsInDirectory_
> and since this method gets called a few hundred times at initialization in
> this project, initialization of Log4J took more than 2 minutes with jar
> packages, compared to a few seconds with class files.
> The performance issue was already mentioned in
> [LOG4J2-184|https://issues.apache.org/jira/browse/LOG4J2-184], only without
> the jar problem.
> The solution to [LOG4J2-175|https://issues.apache.org/jira/browse/LOG4J2-175]
> might cause this, but we are unsure why the methods exactly get called this
> often.
> I just can tell that
> _org.apache.logging.log4j.core.config.plugins.PluginManager#collectPlugins()_
> gets called VERY often via
> _org.apache.logging.log4j.core.config.BaseConfiguration#getPluginManager_,
> which gets called by
> _org.apache.logging.log4j.core.config.XMLConfiguration#constructHierarchy_
> And _collectPlugins()_ is remapped to _collectPlugins(boolean preLoad, final
> String pkgs)_ with _collectPlugins(true, null);_. I'm unsure what this
> preLoad is for, since it gets overridden with false and apparently no kind of
> caching is used. In this method, _resolver.findInPackage(test, pkg);_ takes a
> whole lot of time searching within the jars, as described in the beginning.
> The only way to work around this was for us to extract the custom appender in
> a sperate ant build target, and with this in a seperate jar, to minimize the
> size of the jar, reducing the search time.
> Can you cache things and maybe reduce the calls to _collectPlugins()_? I
> don't see the probability of changing plugins during initialization as
> described in LOG4J2-175 or the comment in
> _org.apache.logging.log4j.core.config.BaseConfiguration#getPluginManager_.
> Introducing a variable for the plugin manager within the for loop in
> _org.apache.logging.log4j.core.config.XMLConfiguration#constructHierarchy_
> might be a good idea!
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
