[
https://issues.apache.org/jira/browse/LOG4J2-1048?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14579725#comment-14579725
]
Gary Gregory commented on LOG4J2-1048:
--------------------------------------
Note that there is a JIRA to redo this feature use the Java 7 file watch
service.
There is also another JIRA to redo all timings using the nanosecond API instead
of the millisecond API.
Nevertheless this is a good catch.
Thank you.
Also note that proposing changes is always better as patch file in unified diff
format.
G
> FileConfigurationMonitor - High CPU usage
> -----------------------------------------
>
> Key: LOG4J2-1048
> URL: https://issues.apache.org/jira/browse/LOG4J2-1048
> Project: Log4j 2
> Issue Type: Bug
> Components: Configurators, Core
> Affects Versions: 2.3
> Reporter: Nikhil
>
> I am observing some high system CPU usage as a result of high number of calls
> to System.currentTimeMillis() in
> FileConfigurationMonitor.checkConfiguration() method.
> The checkConfiguration() method is called as part of Logger's every
> log/logIfEnabled and isEnabled method, therefore getting called thousands of
> times in my application when under moderate load. Each one of these calls
> therefore results in calling System.currentTimeMillis().
> The following code change, where the currentTimeMillis() is called only when
> required seemed to reduce the cpu load
> -final long current = System.currentTimeMillis();-
> -if(((counter.incrementAndGet() & MASK) == 0) && (current >= nextCheck))-
> Fixed as
> {code}
> final long current;
> if (((counter.incrementAndGet() & MASK) == 0) && ((current =
> System.currentTimeMillis()) >= nextCheck)) {
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
