Semaphore issues with zabbix 2.0.2

[c0m4]

Hey

I have a medium installation of zabbix 2.0.2 of about 40 servers. I have one server that the zabbix agent keeps creating more semaphore arrays very quickly up until it blocks everything in it.
I guessing that the zabbix agent just fails to close them and then open new ones because when I stop the agent I can see about the same amount of semaphore arrays(-~5) then when the agent is running. this amount can reach about 100 semaphores within one day.

I've used the following after stopping the zabbix agent service to remove all the leftover semaphore arrays:

# ipcs -s -c | gawk '/zabbix/{system("ipcrm -s "$1)}';

but it is just to keep the server running with zabbix monitoring, my other option is just to stop the zabbix agent altogether, which obviously is not optimal, I need to monitor this server...
in the meanwhile I've increased the amount of allowed semaphores array on the system by running the following command:

# sysctl -w kernel.sem='250	256000	32	1024'

but I'm guessing its just a matter of time until it reach it's new max value.

I would really like to debug and solve this problem but I'm not sure where to start so your help here would be much appreciated. here is some details on this system:
- Centos 5.6
- zabbix agent 2.0.2 (rpm package from http://www.kodai74.net/packages/zabb...rhel/5/x86_64/)

- config:
PidFile=/var/run/zabbix/zabbix_agentd.pid
LogFile=/var/log/zabbix/zabbix_agentd.log
LogFileSize=1
EnableRemoteCommands=1
Server=X.X.0.55,127.0.0.1
ListenIP=X.X.0.32
ServerActive=X.X.0.55,127.0.0.1
Include=/etc/zabbix/zabbix_agentd.d/

(with some userparameters)


- amount of semaphores one the system currently(within one day of running):

# ipcs  -c -s | grep zabbix | wc -l
139

Thanks in advance for your help,

Rudy

[eskytthe]

You can search and report bugs here:
https://support.zabbix.com (ZBX)

There is a guideline here:

https://zabbix.org/wiki/Docs/bug_reporting_guidelines

Hope it helps.
BR
Erik

[c0m4]

actually no.. I did search in the bug reports and couldn't find anything helpful.
I don't want to report a bug since I don't think I have enough data to post for the developers to reproduce...
so I was hoping for some help on solving the problem or instructions on how to extract data from my system in order to solve the problem.