Page 1 of 1

[OM 4646] Too many files open, cannot open any more

Posted: Tue Jan 29, 2008 6:58 am
by Kris
For some time now, I have some users with Outlook problems. I've 150 Premium users, and about 5 users with problems. Problems like Outlook locking up.. weird error messages.. reinstalling Outlook, recreating Outlook profile.. reinstalling and trying different versions of the Scalix connector.. nothing works.

Today, I decided to dig a bit deeper then I did before. First, I started looking at the scalixmail.log file of one of those users with problems. I noticed that there are some messages in there.. mostly, I see "The operation requested is not supported in the current implementation of the service provider". I compared this with a user without problems.. these same message are in that scalixmail.log file as well.. Okay.. digging a bit deeper in the logfile.

Code: Select all

A fatal error has occurred - see the system error log
Too many Files open. Cannot open any more


Okay.. that doesn't sound good.. so I started digging at the server.. and I noticed that there are a lot of those [OM 4646] errors in the fatal logfile.. Most of them from users that complain a lot about Outlook trouble. I also noticed that this error shows up after 3th October 2007. And that's a couple of days after I replaced my RHEL3/Scalix 10 server, with a CentOS5/Scalix 11 server.

So, apparently I'm reaching my max files open limit. But why always the same users with these problems?

Code: Select all

cat  /proc/sys/fs/file-max
406885


This seems high enough..?

Code: Select all

lsof | wc
44970  402834 4792630


I'm a bit lost.. what should I do to solve this..?

CentOS5
Scalix 11.2
Outlook 2000 with latest Scalix 10 connector (Scalix 11 connector with Outlook 2000 is giving much more problems)

Posted: Tue Jan 29, 2008 9:38 am
by fkienker
Kris you are not alone.

I've seen exactly the same problems with RHEL4, Scalix 11.2, and Outlook 2003. It only appears to happen under heavy traffic from Outlook when the server is under higher than normal load.

It started just after a kernel update to RHEL. No amount of adjusting the maximum number of open files in the kernel parameters resolve it. Rolling back to the older kernel version also did not correct it after is started occurring. Newer kernel updates do not correct it either.

Posted: Tue Jan 29, 2008 10:45 am
by ls-al
This looks like bug 16032 which is fixed with 11.3.

Posted: Tue Jan 29, 2008 10:51 am
by fkienker
Thanks for bringing this to our attention. We had been holding off on 11.3 due to the number of NEW issues it introduced. As they die down we will probably consider moving to it. More so if it fixes this issue

But better the devil you know than the one you don't know!

Thanks again!
Fred

Posted: Wed Jan 30, 2008 6:07 am
by Kris
Ah.. I missed this bug in bugzilla. Good to know that it's fixed in 11.3.. I will try to upgrade as soon as possible..

Thanks!