0

Sometimes I find the AG Cluster role in offline state which results the AG in resolving state. It doesn't allow the application to access the databases during this time. And then after sometime it comes to online state. Why is that happening?

If one of the cluster networks is down, then the cluster role goes to offline state?

2
  • 1
    What does the cluster log tell you about why the resource went offline. Get-ClusterLog docs.microsoft.com/en-us/powershell/module/failoverclusters/… – David Browne - Microsoft Nov 7 '20 at 15:18
  • @DavidBrowne-Microsoft : In Cluster log, it is mentioned as "[RHS - Timeout] Resource 'AG' has not responded to the call LOOKSALIVE:0. The timeout to respond has been exceeded by 15 milliseconds, taking recovery actions". Prior to that, "[RES] Network Name <AG_listener>: AccountAD: CheckIfPasswordIsInSync for listener returned 0" is present. – Sree Lekha Nov 8 '20 at 3:01
0

Seems it's your follow-up question of this.. Hope you got root cause of the issue using FailoverDetection tool. Perhaps the cause could be from either Internal Health Issue or system wide issues as detailed here...

Why is that happening?

As you found in error log, the AG trying to take recovery actions, as part of that either AG resources could fail-over to new primary replica as AG is determined unhealthy and will report failures on IsAlive calls, or if the quorum lost, resources goes offline.

Further reading on lease-time and health-check timeout

If one of the cluster networks is down, then the cluster role goes to offline state?

Yes, if the quorum lost.
For instance, if the WSFC formed with 2 nodes (total 2 votes), to have quorum to be healthy, always more than half of the quorum vote nodes to be alive. In this case if one node is down cluster resource may go off-line as the cluster consider to be unhealthy while there is 1 vote is not more than half of the total votes.

To maintain healthy-cluster, configure a file-share-witness (Node and File Share Majority), so that the quorum can have an additional vote from file-share to support the cluster to be healthy. If one node is down there will be 2 valid votes (1 Node + 1 File-share) out of 3.

In Summary:

  • If you found that the root cause is quorum lost, you may need to configure proper quorum mode.
  • If it's Network interface failure, consider having dedicated NIC for the cluster heart-beat communication (screenshot for your reference).
  • If the cause is aside from those two, you need further investigate to find bottlenecks of resource pressure in servers: some free-tools from Microsoft (Ref1, Ref2, Ref3), or third-party tools can help.

Multiple NIC for WSFC: enter image description here

2
  • Thanks! Please help me here. If windows fails to install update related to Microsoft Defender Antivirus antimalware platform - KB4052623. Does it have any affect on the cluster or AG? – Sree Lekha Nov 9 '20 at 17:11
  • @SreeLekha, I cannot confirm that, but if you sure that the issue noticed after the particular windows update, then you need to roll back it (for the confirmation). In addition, be sure applying the exclusions... – Shekar Kola Nov 11 '20 at 9:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.