ESX 6.5 - Configuring Automatic Storage Rescan?

[bertram]

Hi,

I'd read the technical paper (https://www.starwindsoftware.com/resour ... sphere-6-5) because it's the scenario I want to use with a 2-node (compute & storage seperated) SAN, active/active.

While reading the part with the powershell script that initiates storage rescans on the esx hosts my first thought was that this must be a failback-failback-... solution if something goes wrong anyway.

But after a few tests I realized that it's crucial to rescan the storage nodes on the esx hosts to get the iSCSI path back from offline/failed to active if a Starwind node was down for any reasons (e.g. reboot). WTF?

I've never had this strange behaviour with any iscsi solution and VMware before. All iSCSI targets from other vendors always became recognized and automatically activated if they failed for some reasons for a period of time.

Why is the rescan necessary with the Starwind solution?

[Boris (staff)]

With Compute&Storage Separated scenario, no storage rescan script deployment is necessary.
With Hyperconverged scenario it is required as sometimes VMware fails to recover the previously failed path even if it is available again and would not continue attempting to recover it until storage rescan is triggered explicitly. This is not any kind of behavior that StarWind has got, it's about the way VMware behaves.

[bertram]

Thanks for the clarification. Maybe a remark with a short explanation in that technical paper would help somewhat.

Aynway, I've tried some settings and found that it's an issue(?) with the CHAP settings in VMware. Because I've set a global CHAP challange for the initiators in Starwind I've set "CHAP required" in VMware. But for any reasons this leads to the behaviour I posted above. When I set "CHAP prefered" the path become active again automatically as expected.

[Boris (staff)]

Thank you for the suggestion. We will consider adding some sort of explanation into the tech paper.