Challenge
When nworks uses "Refresh Hardware" calls, hardware status is not updated on vSphere 5
Cause
vSphere 5 does not update the hardware status data available through API - this can be verified in vSphere Managed Objects Browser, which shows an outdated status, while the host MOB will show the correct status. By default, this information is not updated in vCenter 5. The VI Client, however, has the valid data, since it uses CIM service for watching the sensor readings.
Solution
Solution available with nworks 5.7: ensure CIMXML is working properly. Refer to KB1568 for more detail.
Workaround provided by VMware:
To avoid the discrepancy between VC MOB and ESX MOB, you have to configure "throttleinterval" (in seconds) in \etc\vmware\vpxa\vpxa.cfg file (found on the host) as below:
<config>
....
<vpxa>
....
....
<healthSystem>
<throttleInterval>1200</throttleInterval>
</healthSystem>
</vpxa>
<workingDir>/var/log/vmware/vpx</workingDir>
Then run the below command to restart vpxa:
services.sh restart vpxa
No other services need to be restarted.
More Information
Applying the workaround causes significant amount of host syncs at idle time and affects VC scalability, so it is not recommended to configure very frequent updates.