Partner HAImage "Not Synchronized" after One Node is Down & Full Synchronization

[muhlitfan]

Hello, I want to ask some question regarding the synchronization of VSAN Starwind. I'm using VSAN Starwind Free and currently using 2 Node HAImage and another node as the witness using SMB Witness for node majority failover strategy. This question is for when I want to test failover after all the HAImage in all node is Synchronized.

1. When one node is down, the other node (partner node) turns into "Not Synchronized" state. Is this normal? What's the cause of this behaviour? Is losing the connection to the SMB Witness node could be the one causing this?
2. Related to the first question, after some tweaks with the sync traffic share. That behavior never happened again after I've tested and set the sync traffic share to 30% or 40%. Why is this the case? And what does Client Request Processing actually really mean?
3. Based on this knowledge base https://knowledgebase.starwindsoftware. ... may-start/. Is this only a recommendation? Is it still possible Full Sync to occur when I've followed all the instructions from there?

[yaroslav (staff)]

1. When one node is down, the other node (partner node) turns into "Not Synchronized" state. Is this normal? What's the cause of this behaviour? Is losing the connection to the SMB Witness node could be the one causing this?

The affected node, i.e, the one you take down, needs to go not synchronized; the other says PARTNER is not synchronized. You can see the iSCSI initiator. If the storage is there, all good.
Node majority Failover Strategy does not take the condition where the node is isolated. That node is marked as "not synchronized". If one node is down and the Witness is down or the connection to it is interrupted, the only synchronized node will go "not synchronized," and marking the HA device as synchronized will be needed.

2. Related to the first question, after some tweaks with the sync traffic share. That behavior never happened again after I've tested and set the sync traffic share to 30% or 40%. Why is this the case? And what does Client Request Processing actually mean?

Sync share does not affect replication stability.
Client request processing implies VMs or other client connections to StarWind VSAN. In other words, anything that uses data.

3. Based on this knowledge base https://knowledgebase.starwindsoftware. ... may-start/. Is this only a recommendation? Is it still possible Full Sync to occur when I've followed all the instructions from there?

I don't follow. What recommendation are you referring to?

[muhlitfan]

I don't follow. What recommendation are you referring to?

Nevermind yaroslav, I've actually misunderstood some points from the knowledge base I refer. I've done my HAImage setup to not use Write-Back cache and use disk-based synchronization journal. But it seems I miss the point that say "If the Partner that should be the source of synchronization gained “Synchronized” state after initiating “Mark as synchronized” command". I've actually recently force "mark as synchronized" the source synchronization of the HAImage using Powershell, and now it's going into full synchronize.

Thank you btw for your response. Have a nice day!

[yaroslav (staff)]

You are always welcome. Sorry to read that there was an accident, but I am glad to hear that things work well.
Wishing you a nice day!