VSAN Datastore Size not reflecting actual size

[kirand]

Hello everyone,

I have configured VSAN Cluster with 3 nodes and VSAN datastore size is 6TB,  When i disconnect one of the node from cluster, datastore size in WebClient still shows 6 TB. But, if put the host into maintenance mode and remove from cluster, datastore size reduces to 2 TB. Actual Datastore size is not reflecting in WebClient when one of the host is removed/disconnected without entering into maintenance mode.

Is this expected behavior? or is there anything we should be doing after a host gets disconnected from cluster to display the correct VSAN datastore size.

[depping]

Well when you disconnect them they are not taken out of the cluster, they are just disconnected from vCenter.

[kirand]

In a four node cluster, when one node is taken out, doesn't VSAN recalculate the datastore size and reflect properly? If plan to continue without replacing the failing node, the datastore size will mislead for capacity estimation while creating VMs. Please share your thoughts.

[depping]

kirand wrote:

In a four node cluster, when one node is taken out, doesn't VSAN recalculate the datastore size and reflect properly? If plan to continue without replacing the failing node, the datastore size will mislead for capacity estimation while creating VMs. Please share your thoughts.

Yes if you "leave" the cluster by placing it in maintenance mode or issuing a "leave" command on the esxi console it will. Not when you disconnect a server.

[kirand]

Thanks for explanation, but i believe this needs to be addressed in upcoming releases.

I have a VSAN Cluster with N nodes and one of the host failed (node failure, not manually removed), and I am planning to continue (with n-1 nodes) with out replacing the failed node. In this case, datastore size reflecting in WebClient is misleading me for my VM capacity estimation.

[joergriether]

Can someone at vmware please jump in and clarify: Is there really no way at all to "force" a global vsan-sizing-recalculation, assuming a host suddenly fails and permanently (assuming is crashed and will never come back)?

Best regards,
Joerg

[depping]

joergriether wrote:

Can someone at vmware please jump in and clarify: Is there really no way at all to "force" a global vsan-sizing-recalculation, assuming a host suddenly fails and permanently (assuming is crashed and will never come back)?

Best regards,
Joerg

will try to validate this in my lab today, if not possible I will contact engineering and file a feature request / bug report.