[solved] Multiple SNMP Get Timeout for Host

[mbails]

Operating System: Windows 2008
Webserver: IIS
Cacti:0.8.7g
Spine: 0.8.7g
MySQL:5.1.50
PHP:5.2.14
RRDTool (Cygwin or Win32 version):1.2.30
Net-SNMP:5.5.0.2
Cygwin (cygwin1.dll version): 1.7.5
Plugin Architecture: 2.9 beta2
Plugins: Aggregate, Settings, RrdClean, Thold

I've recently upgrade a server to 0.8.7g and PIA 2.8 then started running into:

11/17/2010 12:25:47 PM - SYSTEM THOLD STATS: Time:1.9117 Tholds:658 Hosts:0
11/17/2010 12:25:45 PM - SYSTEM STATS: Time:45.6420 Method:spine Processes:2 Threads:15 Hosts:80 HostsPerProcess:40 DataSources:3899 RRDsProcessed:2167
11/17/2010 12:25:25 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host1', and OID:'.1.3.6.1.2.1.25.2.3.1.5.3'
11/17/2010 12:25:22 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host1', and OID:'.1.3.6.1.2.1.25.2.3.1.6.3'
11/17/2010 12:25:19 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host2', and OID:'.1.3.6.1.2.1.25.2.3.1.5.5'
11/17/2010 12:25:17 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host3', and OID:'.1.3.6.1.2.1.25.2.3.1.5.2'
11/17/2010 12:25:16 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host2', and OID:'.1.3.6.1.2.1.25.2.3.1.6.5'
11/17/2010 12:25:15 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host4', and OID:'.1.3.6.1.2.1.25.2.3.1.6.2'
11/17/2010 12:25:15 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host3', and OID:'.1.3.6.1.2.1.25.2.3.1.6.2'
11/17/2010 12:25:13 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.5.35'
11/17/2010 12:25:13 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.6.35'
11/17/2010 12:25:11 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.5.34'
11/17/2010 12:25:11 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.6.34'
11/17/2010 12:25:11 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.5.31'
11/17/2010 12:25:11 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'host5', and OID:'.1.3.6.1.2.1.25.2.3.1.6.31'

I've since done a clean install on a new server then migrated over the database, RRAs, resources, etc. Everything is running fine except I'm still receiving the errors above.

I need some help figuring out how to troubleshoot this. I've searched the forum and haven't found anything relevant.

[mbails]

I've narrowed it down to the Host MIB - Available Disk Space Graph and Data Source. It's just timing out.

Edit: Disregard. I was wrong.

[Ilya Evseev]

The same problem - multiple messages in cacti.log:

Code: Select all

08/02/2011 05:00:23 PM - CMDPHP: Poller[0] WARNING: SNMP Get Timeout for Host:'winserv.lan', and OID:'.1.3.6.1.2.1.25.2.3.1.6.3'

Changing properties of Windows-based host in Cacti devices list:
a) increasing "SNMP timeout" from 500ms to 2s,
b) or decreasing "Maximum OID's Per Get Request" from 10 to 5...
...doesn't help.

[BSOD2600]

Try a timeout of 5 seconds... yes, I know its ridiculously high.

[Ilya Evseev]

Try a timeout of 5 seconds... yes, I know its ridiculously high.

Does not help.

May be standard M$ SNMP agent can be replaced by something more reliable?
I've tried ADMLink SNMP Agent from http://rcolonel.tripod.com -- it works very strange.
What about http://marksw.com/snmpv3agent/windowsagent.html ?

[BSOD2600]

I haven't tried any other snmp agent.

if you snmpwalk .1.3.6.1.2.1.25.2.3.1.6.3 for that problem device, what do you get?

[Ilya Evseev]

if you snmpwalk .1.3.6.1.2.1.25.2.3.1.6.3 for that problem device, what do you get?

.1.3.6.1.2.1.25.2.3.1.6.3 = INTEGER: 143675805
...or "Timeout: No Response from winhost.lan".

[BSOD2600]

For that OID, the device gives you a number or times out? Sounds like either a problem with your device being flaky or other network problems which are not related to cacti.

[Ilya Evseev]

For that OID, the device gives you a number or times out?

Sometimes a number, more often timeout.

Sounds like either a problem with your device being flaky

Perfmon on Winhost(s) reports no CPU nor network problems.

or other network problems which are not related to cacti.

Linux- and BSD-based neighbors of Winhost(s) are accessible normally.

This is common problem for all winhosts in all our networks.
Can it be inspired by buggy Microsoft SNMP agent?

[BSOD2600]

Ah, right I forgot this OID was for hrStorageUsed. Yes, the microsoft implementation of querying disks is VERY laggy for some reason. You should be setting the snmp timeout to 3 - 5 seconds.

[Ilya Evseev]

You should be setting the snmp timeout to 3 - 5 seconds.

Already tried without positive effect:
http://forums.cacti.net/viewtopic.php?f ... 26#p215426

[BSOD2600]

Well then your host is slower than average. Increase the snmp timeout to 10-15 seconds.