Hi Rakesh! I can't be sure what your error really is, that was just a guess.
However, if changing the ulimit solves the problem, it's a quick fix, and easy to test. In theory, the issue can "pop up" anytime on a system where the number of file descriptors is set too restrictively, since the number of files opened by a DB like Derby will increase with the amount of content stored. So if the setting for files is "too tight", adding some content or performing upgrades (which changes many nodes and presumably uses many temporary files) can cause the error. Regards from Vienna, Richard Von: [email protected] [mailto:[email protected]] Im Auftrag von Rakesh Vidyadharan Gesendet: Mittwoch, 04. Mai 2011 16:09 An: Magnolia User-List Betreff: Re: [magnolia-user] AW: Too many open files with 4.4.3 Hi Richard, We are using H2, but it may have similar issues as Derby. However, what would suddenly cause this issue with 4.4.3, since we do not experience it after reverting back to 4.4.2? I tried it a few times on my development environment with 4.4.3 and the error is very repeatable. Thanks Rakesh From: "Unger, Richard" <[email protected]<mailto:[email protected]>> Reply-To: Magnolia User-List <[email protected]<mailto:[email protected]>> Date: Wed, 4 May 2011 07:39:30 -0500 To: Magnolia User-List <[email protected]<mailto:[email protected]>> Subject: [magnolia-user] AW: Too many open files with 4.4.3 Hi Rakesh! Are you using the DerbyDB? "too many open files" is a known issue under Linux with DerbyDB. Solve it by increasing the ulimit for the user running tomcat: ulimit -n 2048 Regards from Vienna, Richard Von: [email protected]<mailto:[email protected]> [mailto:[email protected]] Im Auftrag von Rakesh Vidyadharan Gesendet: Dienstag, 03. Mai 2011 18:44 An: Magnolia User-List Betreff: [magnolia-user] Too many open files with 4.4.3 Hello, We updated to 4.4.3 a few days ago and started noticing the above issue when our scheduled nightly data import and activate job runs. Our job imports and updates nodes in our books workspace from our internal database. We typically activate a few thousand nodes each night after the update process runs. This has so far not been a problem (we have activated around 40,000 nodes using this approach in one run). With 4.4.3, the activation process dies after a few nodes have been activated with the above message. Lsof lists a lot of deleted files and list does not shrink until the JVM is restarted. We reverted back to 4.4.2 and the process is back to normal once again. Anyone faced any similar issues with 4.4.3? Is there any specific cleanup that needs to be performed when using the Syndicator? Thanks Rakesh ________________________________ ---------------------------------------------------------------- For list details see http://www.magnolia-cms.com/home/community/mailing-lists.html To unsubscribe, E-mail to: <[email protected]<mailto:[email protected]>> ---------------------------------------------------------------- ________________________________ ---------------------------------------------------------------- For list details see http://www.magnolia-cms.com/home/community/mailing-lists.html To unsubscribe, E-mail to: <[email protected]<mailto:[email protected]>> ---------------------------------------------------------------- ________________________________ ---------------------------------------------------------------- For list details see http://www.magnolia-cms.com/home/community/mailing-lists.html To unsubscribe, E-mail to: <[email protected]<mailto:[email protected]>> ---------------------------------------------------------------- ---------------------------------------------------------------- For list details see http://www.magnolia-cms.com/home/community/mailing-lists.html To unsubscribe, E-mail to: <[email protected]> ----------------------------------------------------------------
