topic java process at 100% CPU Utalisation in Alfresco Archive

java process at 100% CPU Utalisation

dgildeh — Mon, 13 Jun 2011 02:59:44 GMT

Hi,We upgraded our server from 3.2r to 3.4d last week. During testing we didn't see any issues. However now its in LIVE we've noticed the soffice and java processes seem to be running at 100% CPU utalisation the last few days and kill processes/rebooting/changing JVM settings doesn't appear to make

Re: java process at 100% CPU Utalisation

amitabhandari1 — Mon, 13 Jun 2011 04:36:00 GMT

David,

Are there any messages coming on the console of server. Please try to configure JProfiler which will help in seeking the details about the consumption of memory.

Thanks,

Re: java process at 100% CPU Utalisation

loftux — Mon, 13 Jun 2011 05:16:46 GMT

In that other thread you mentioned running Ubuntu, what version?
Have you updated to the latest version of java JDK and OpenOffice?

Since you are seeing that office.bin hits 100% it can be that it tries to load and convert for indexing a document that is either very large or malformed in some way. So maybe run the latest OpenOffice or LibreOffice downloaded from their respective site - maybe they have better filters. They can be installed alongside the official ubuntu packages.
If you try LibreOffice their will be an error message on startup, but it will work fine after that anyway.

Try adding these log statements
log4j.logger.org.alfresco.util.exec.RuntimeExec=debug
log4j.logger.org.alfresco.repo.content.transform.ContentTransformerRegistry=debug
log4j.logger.org.alfresco.util.OpenOfficeConnectionTester=debug

Maybe that can give what document cause trouble.

Re: java process at 100% CPU Utalisation

dgildeh — Mon, 13 Jun 2011 09:56:08 GMT

Thanks to both of you for your quick response. I tried for ages to get JConsole to work remotely (the servers running on AWS and for whatever reason I couldn't). A couple of hours after I posted this, the JVM crashed again after I uploaded a PDF. Looking at the JVM Log it showed PDFBox to be the root of the cause and I found this issue on Alfresco Jira: http://issues.alfresco.com/jira/browse/ALF-8097

It looks like this was the root cause of the server crashing several times the other day, which I noticed after I tested uploading a PDF and the server crashed everytime I tried to view the document thumbnail. I updated the libraries and the crashing stopped. It also looks like after a few hours the CPU has returned to 0% utalisation when noone is using it since this change, leading me to suggest that this error was screwing up the background indexing process as well.

It looks liike updating the PDFBox libraries has resolved both the crash issue and CPU utalisation issue, so I'm cautiously optimistic that this is resolve.

Thanks again,

David