Zimbra Timeout external ldap issues

[csf-clacroix]

Similar to : Zimbra login timeouts

We have problems where mostly pop and imap services start to get slow, disconnecting and cause our users to call in the call center to report the issue. Up to now the webmail is not affected by this issue. Most of our users are lookout/lookout express users but problem is there in Mail and ThunderBird.

I've been searching around and i havn't found much information on this.

Being new to Zimbra, i started looking around for bits of information and it's helping me find my way around the system.

We run:
Novell eDir8.8 sp2 as our external ldap

Zimbra:
Release 6.0.7_GA_2473.RHEL5_64_20100616215237 RHEL5_64 FOSS edition

Load is always under 1 on all my Zimbra servers
I have plenty of free RAM and it's not touching Swap
CPU is under used even on the MTA doing all of cpu intensive filtering.
Everything is connetcted with Gig ethernet.
No errors on the net interface
No disk errors.

One thing i noticed see that both my backend servers ( the ones running imap, pop and webmail ) are connecting to our external ldap server for authentication. This fine and it's how we want it to be but the connexion pool just keep on growing until people start calling with issues on their pop/imap clients. I noticed that problem start to occure on a backend server when i have over 1200 concurent ldap connection on it. The connection pool just seam to grow and it never goes down.

I'm watching it with a small shell script appending to a file and it seams to just grow until i zmcontrol restart on the backends.

----
date > log
/usr/bin/lsof -i -n -P |/bin/grep :389|/bin/grep 105.168|wc -l > log
/usr/bin/lsof -i -n -P |/bin/grep :389|/bin/grep 105.169|wc -l > log
----

Thanks

[csf-clacroix]

any idea anyone?