agent stops monitoring on W2003

[agraaff]

For some reason the Zabbix agent stops collecting data on some of my Windows2003 servers.
after a restart of the agent all is fine again for 10 min, a hour, a few hours, a day... and then it stops collecting again.
usualy after a few hours.

had it in the 2.04 agent, 2.09 agent and still in the 2.2.0 agent.

below some logging from the agent.

6112:20131125:100428.806 Starting Zabbix Agent [vm00367.asp-services.net]. Zabbix 2.2.0 (revision 40147).
6112:20131125:100428.806 using configuration file: c:\program files\zabbix\zabbix_agentd.conf
6004:20131125:100428.806 agent #0 started [collector]
4624:20131125:100428.837 agent #1 started[listener #1]
2608:20131125:100428.837 agent #2 started[listener #2]
4160:20131125:100428.837 agent #3 started[listener #3]
4624:20131125:100934.490 cannot get required buffer size for counter path 'Process(*)\Working Set': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:100942.083 cannot get required buffer size for counter path '\Process(*)Thread Count': [0xC0000BC4] The specified counter path could not be interpreted.
4160:20131125:100947.193 cannot get required buffer size for counter path '\\Memory\Pool Nonpaged Bytes': [0xC0000BC4] The specified counter path could not be interpreted.
4160:20131125:101921.779 cannot get required buffer size for counter path 'Process(*)\Working Set': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:101929.638 cannot get required buffer size for counter path '\Process(*)Thread Count': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:101935.013 cannot get required buffer size for counter path '\\Memory\Pool Nonpaged Bytes': [0xC0000BC4] The specified counter path could not be interpreted.
4624:20131125:102950.443 cannot get required buffer size for counter path 'Process(*)\Working Set': [0xC0000BC4] The specified counter path could not be interpreted.
4624:20131125:102952.615 cannot get required buffer size for counter path '\Process(*)Thread Count': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:102954.115 cannot get required buffer size for counter path '\\Memory\Pool Nonpaged Bytes': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:103957.904 cannot get required buffer size for counter path 'Process(*)\Working Set': [0xC0000BC4] The specified counter path could not be interpreted.
4624:20131125:104000.060 cannot get required buffer size for counter path '\Process(*)Thread Count': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:104001.701 cannot get required buffer size for counter path '\\Memory\Pool Nonpaged Bytes': [0xC0000BC4] The specified counter path could not be interpreted.
4624:20131125:105009.318 cannot get required buffer size for counter path 'Process(*)\Working Set': [0xC0000BC4] The specified counter path could not be interpreted.
4624:20131125:105011.271 cannot get required buffer size for counter path '\Process(*)Thread Count': [0xC0000BC4] The specified counter path could not be interpreted.
2608:20131125:105012.912 cannot get required buffer size for counter path '\\Memory\Pool Nonpaged Bytes': [0xC0000BC4] The specified counter path could not be interpreted.
.
.
.
.
2608:20131125:185052.223 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
4624:20131125:185155.254 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
2608:20131125:185313.378 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
4624:20131125:185331.315 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
2608:20131125:185413.377 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
4624:20131125:185431.314 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
2608:20131125:185513.376 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]
4624:20131125:185531.313 PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text [0x00000057]

[Mython]

I´m having the same issue in one of my windows 2003 server...

Agent 2.2.1

PERF_COUNTER(): call to PdhOpenQuery() failed: [0x00000102] unable to find message text: [0x0000013D] unable to find message text

Any idea to solve this??...

[ingus.vilnis]

Hello!

Does the Zabbix agent on your Windows server crash (not responding) or it simply stops sending data?

What performance Items are you monitoring on that host?

Best Regards,
Ingus

[Mython]

Hi Ingus,

Well... for example the default template of the OS Windows...
perf_counter[\234(_Total)\1402]
perf_counter[\234(_Total)\1404]
perf_counter[\2\16]
perf_counter[\2\18]
vm.memory.size[free]
system.swap.size[,free]
agent.hostname
proc.num[]
perf_counter[\2\250]
system.cpu.load[percpu,avg1]
system.cpu.load[percpu,avg5]
system.cpu.load[percpu,avg15]
system.uname
system.uptime
vm.memory.size[total]
system.swap.size[,total]
agent.version

among others..

And yes, the agent just stop sending data, because the agent service is up.(not crashing...)
thanks in advance.

[ingus.vilnis]

Ok, I have to start this post with a huge disclaimer. The following suggestions are purely a guess, I have not tried them myself, I don't know how it works, nor how it is related to your issue. If you have a production server there (of course you have, I cannot imagine someone keeping W2003 server at home just for fun), you can try to do this at your very own risk, and before that do a serious investigation on what you are trying to do. So now, after you cannot blame me, I have found some pointers for you.

• Download PHD.dll file (http://support.microsoft.com/kb/284996/en)
• Put this file in the same directory where your zabbix_agentd.exe is
• Run lodctr /R to rebuild the performance counter tree (http://technet.microsoft.com/en-us/l.../bb490926.aspx)
• Restart zabbix_agentd.exe

Once again, I don't know if this can help. Any comments from experienced Windows admins would be highly appreciated here.

Best Regards,
Ingus

[Mython]

Thanks for your help, since I do not have how to check quickly if this is a solution, (after a restart the agent all is fine again for 10 min, a hour, a few hours, a day... and then it stops collecting again)
I will let you know as soon as a better behaviour in the agent is noted (one or two days).

Thanks again in advance.

[agraaff]

Yes they are production servers
The agents keeps running but just stops responding.

Funny, also thougt it might be some error in the perf counters so i tried your sugestion of rebuilding the performance counters already. Although i was sceptic at the time because all the w2003 servers stop monitoring at some point (some within the hour some after a few days and the rest in between) but perfmon keeps working fine when loggong on to the server.
It didn't make a difference though.

[ingus.vilnis]

Hi agraaff,

Does the zabbix_agentd still crash after rebuilding the performance counters?
Did you try anything with the PDH.dll file as well?

Best Regards,
Ingus