The HostMonitor Agent, the VCShm Resource, and the VCShmg Service Group

  • Article ID:100019873
  • Modified Date:
  • Product(s):


The HostMonitor Agent, the VCShm Resource, and the VCShmg Service Group


Notes on HostMonitor, the VCShm resource, and the VCShmg group:

Per design, VCS 5.0 MP3 automatically starts the HostMonitor agent.  VCS creates a "hidden" resource called VCShm, and puts it into a "hidden" service group called VCShmg.  These resources and groups DO NOT show up with generic "hastatus -sum", "hagrp -display", or "hares -display" commands.  However, if you specify them on the command line you can view them:

To view the HostMonitor type:
# hatype -display HostMonitor

To view the VCShm resource:
# hares -display VCShm

To view the VCShmg Service Group:
# hagrp -display VCShmg

The CPUUsageMonitoring attribute controls what action (if any) VCS takes based on the VCShm resources findings.  The agent will still run when CPUUsageMonitoring is set to 0, and may still log messages in the engine log.  However, VCS will not take any action based on its findings.

Please refer to the VCS Users Guide for more information on CPUUsageMonitoring.

The HostMonitor agent will log a message in the engine log when either CPU or Swap utilization increases beyond 60% (info), 70% (warning), 80% (error) or 90% (critical). Decreases below these thresholds are not logged.  The HostMonitor agent will continue to log one message every two hours as long as the utilization stays at its current high (above 60%) value.

Collecting Additional Performance Statistics/logging
It may be useful to log additional performance statics.  There are two ways to do this:

To collect additional CPU / Swap statistics from _had_ in /var/VRTSvcs/log/engine_A.log:
# halog -addtags DBG_HOSTMON

To collect additional CPU / Swap statistics from the HostMonitor agent into /var/VRTSvcs/log/HostMonitor_A.log:
# haconf -makerw
# hatype -modify HostMonitor LogDbg DBG_3 DBG_AGINFO
# haconf -dump -makero

Removing additional logging
# halog -deltags DBG_HOSTMON

# haconf -makerw
# hatype -modify HostMonitor LogDbg -delete -keys
# haconf -dump -makero

Correlating Usage reported by HostMonitor with that of other tools
Each platform has a different command against which you can compare the usage reported by the HostMonitor agent:

# sar -P ALL 1 10

# sar -M -u 1 10

Compare the usage reported by HostMonitor with the info found in the /proc/stat file

#mpstat 1 10

Prevent reporting by HostMonitor
The HostMonitor thresholds are not configurable.  

To prevent the agent from starting.
  • Removing the execute permission from the /opt/VRTSvcs/bin/HostMonitor binary.
  • Move the /opt/VRTSvcs/bin/HostMonitor file to a different file name.
  • Restart HAD on all the nodes in the cluster (hastop -all -force)

To prevent the agent from reporting and to temporarily remove the resource and group.
  • /opt/VRTSvcs/bin/haagent -stop HostMonitor -force -sys <node>
  • Remove agent resource and group
            haconf -makerw
            hares -delete VCShm
            hagrp -delete VCShmg
            haconf -dump -makero

Note: Removing the resource and group will only be temporary. If you stop and start had the group and resource will automatically be rebuilt.

Related Articles

"VCS CRITICAL V-16-1-50086 CPU usage on hostname is 100%" on low usage Linux VCS nodes

VCS INFO V-16-1-50086 CPU usage on &lt;nodename&gt; is &lt;percent&gt;%

Was this content helpful?

Get Support