SRM - Frequent Recomputation of Device Groups & Datastore Groups
Article ID: 312722
Updated On:
Products
Issue/Introduction
1. Upgraded vCenter server is shows very frequent recompute device groups / datastore groups (every 20-30 seconds) from SRM.
2. SRM has cascading logging behavior to vCenter and may fail to run test recovery or recovery correctly.
vpxd.log:
warning vpxd[34575] [Originator@6876 sub=AuthorizeManager opID=ff8566fa-6ba8-4774-a9ed-b2d7aebb4be0-125068-80] Refresh function is not configured.User data can't be added
Environment
Cause
In the logs we can see excessive events occurring related to vCLS can be seen in vpxd logs. This is a known issue and was resolved by multiple customers following the retreat action as per the link.
Resolution
Putting a Cluster in Retreat Mode
Put all clusters connected in the vCenter with the issue into retreat mode. Here is a snippet of powercli script
$CluMoref = $Cluster.ExtensionData.MoRef.Value
$vclsname = "config.vcls.clusters.$CluMoRef.enabled"
# add the setting first
New-AdvancedSetting -Entity $VCSA -Name $vclsname -Value $true -Confirm:$false # creates the adv stg, to be disabled as part of a shutdown script
# once created, it can be changed to $false - this effectively removes the vCLS agent VMs (but also affects DRS/HA, so only done during shutdown
Get-AdvancedSetting -Entity $VCSA -Name $vclsname | Set-AdvancedSetting -Value $false -Confirm:$false
# then add the reverse to the startup script - this creates the vCLS agent VMs again
Get-AdvancedSetting -Entity $VCSA -Name $vclsname | Set-AdvancedSetting -Value $true -Confirm:$false
Workaround:
Retreat vCLS and enable vCLS for all clusters.
Additional Information
After the Upgrade from Vcenter 7.0 U2 to U3 the three vSphere Cluster Services (vCLS) VMs are gone.
Impact/Risks:
The only impact will be that until the vCLS is back online, DRS will be offline. These actions must be done during a maintenance window.
Feedback
