[Zope-dev] running out of available tcp ports on a zeo cluster

[sathya]

Hello folks,

We are testing a zeo cluster in a vmware environment (zeo clients are running on multiple windows 2003 server guest vms). 
We are using Apache as a load balancer using mod_rewrite (on a separate windows 2003 guest vm ).
We are using load runner to generate load 
After hitting about 40 concurrent users we are getting tcp errors on apache  indicating that there are no available ports for accepting connections.
It seems like the connections are hanging on for a longer period of time causing the available ports to get exhausted. The errors get worse on increasing the concurrent users. It seems as though there is a problem  recycling ports for new connections.
Has anybody seen this problem before ? Is there an issue with connections between apache and zope not closing down ?

We have run similar tests on linux before and have not faced this problem ( I know, but the client wants windows :(  )
 

Any feedback will be greatly appreciated.

Regards
sathya
-- 
ZeOmega LLC
Integrated Care Management software
Frisco Tx
214-618-9880
www.zeomega.com
--