Doubts about HA operation

[chendin]

I have configured StarWind Virtual vSAN® 2-node Hyperconverged Scenario and verifying the operation, when there is a node down, it makes a full recovery and during the recovery time
the ISCSI storage unit(s) are not accessible.
Is this normal?
In other vSan solutions that I have tested, in case of a synchronization problem, a master node is chosen, which will become accessible, while the other one is rebuilt.
Is this possible in StarWind Virtual vSAN free?

I have also seen that in StarWind you usually create several ISCSI units while in other solutions you create 2 units, one with little capacity and one with full capacity (to create the ha in Vcenter).
Is it worth to create several units? better one with all the space?

Thank you very much

[yaroslav (staff)]

Welcome to StarWind Forum.

In other vSan solutions that I have tested, in case of a synchronization problem, a master node is chosen, which will become accessible, while the other one is rebuilt.

StarWind VSAN features active-active replication. You can see priority on each device, but those serve for situations where the synchronization link is lost. See more at viewtopic.php?f=5&t=5731.
Storage can be lost only if there is misconfiguration. Please make sure the storage is connected from both nodes and replicated before and at least one node stays synchronized during the test. More on how to restart the node https://knowledgebase.starwindsoftware. ... installed/.
Make sure your deployment aligns with best practices https://www.starwindsoftware.com/best-p ... practices/ and system requirements https://www.starwindsoftware.com/system-requirements

I have also seen that in StarWind you usually create several ISCSI units while in other solutions you create 2 units, one with little capacity and one with full capacity (to create the ha in Vcenter).

You can create up to 64 targets. It can be one device, it can be several devices (e.g., a datastore per VM, or group of VMs). I'd suggest creating multiple devices to balance i/o between multiple targets. Also, smaller devices make full synchronization granular (if interrupted, you have less capacity to resync).

[chendin]

Thank you very much for the quick answer.
I don't explain myself very well I think,
What I mean is that while it is synchronizing I do not have the ISCSI storage accessible in the vcenter
Could it be a configuration problem? If so, what configuration should I set? or if we configure priority it will work correctly

[yaroslav (staff)]

No worries. I am always happy to help.

What I mean is that while it is synchronizing I do not have the ISCSI storage accessible in the vcenter

The paths to not synchronized node should be missing. The synchronized node should be accessible over iSCSI.

Could it be a configuration problem? If so, what configuration should I set?

Please check if the storage is connected from both nodes to both ESXi hosts. If you have more than two ESXi hosts, make sure they are talking to both nodes too.
If the storage disappears completely upon the restart, there is a risk that the path is missing due to not being rescanned properly.
In CVM there is an ongoing rescan issue now which is going to be addressed in the next build (let me know if you are interested in the manual workaround so I can prepare it for you).
In a nutshell, make sure you have paths up to all storage nodes before restarts.