Monday, April 2, 2012

SCOM R2: EventID 2023.

Bumped into an issue where the OpsMgr event log of an OpsMgr Agent logged EventID 2023: ’…The health service has removed some items from the send queue for management group since it exceeded the maximum allowed size of xyz megabytes…’. This event was shown once per hour.

Cause
As it turned out, this server ran many monitored objects, services and applications. As a matter of fact, too many. Normally the queue for a SCOM Agent is set to 15 megabytes (15360 KB).

Solution
The maximum queue size is set to 75 megabytes (76800 KB). Afterwards the Agent is restarted and all is well again. The maximum queue size is 256 megabytes (262144).

The register key you need to update is to be found here: HKLM\SYSTEM\CurrentControlSet\services\HealthService\Parameters\Management Groups\<MG NAME>\MaximumQueueSizeKb.

Advice
Whenever you bump into a server where there are really many objects to be monitored and not all the collected information seems to reach SCOM, try these steps as described here:
image

This applies to SCOM R2 Agents and OM12 Agents as well.

1 comment:

Thief_ said...

The way I found this issue was that SCOM was sending out notifications like this:

Alert: Health Service Heartbeat Failure
Source: myserver.mydomain.com
Path: Microsoft.SystemCenter.AgentWatchersGroup
Last modified by: System
Last modified time: 11/14/2012 7:58:52 AM Alert description: The System Center Management Health Service on computer myserver.mydomain.com failed to heartbeat.

The solution presented here fixed my issue immediately in SCOM 2012 CU1. I now have a whole bunch of working agents as they should be.

Thanks so much for posting!