EVA4400 - Vdisk deletion

[Dinesh Pahari]

Hello Everyone,

Background:
A RAID0 Vdisk was created for some testing purpose. That Vdisk had a CA enabled to DR array as well.

Problem summary:
A disk failed. Obviously, the EVA experienced problem repairing the RAID0 Vdisk. Also the failed disk was not ungroupped and was still on the disk group and was inaccessible.

Action taken:
Because the Vdisk didn't have any data on it, we opted to delete the DR group and the Vdisk. when we pressed the button to delete the DR group, the EVA hung and caused an outage. We had to reboot the controllers to get the storage back online.
After the array came back online, we could access the failed disk but it was still in the disk group. As per HP's advise, we removed the disk and replaced with new one. Everything then went OK with the failed disk.

But we still having problem with deleting the DR group related to that RAID0 Vdisk. HP engineering is involved and looking into it but just wanted to know if someone had this issue before and if there is known solution.

Thank you with anticipation.

Dinesh Pahari

[Víctor Cespón]

Yes, seen the same issue here
No, no solution

After the disk failure you get the dialog box saying that a disk group failure has occurred and some vdisks have failed.
Then you have the options to continue with no changes / start deletion process / use vdisk as-is.

The theory is that start deletion process will remove the VRAID 0 vdisk, but if it's part of a DR Group, it does not work.

Trying to delete the DR Group crashes the controllers with TC:0x075b0100 "An I/O that should NOT fail, did"

Then you get a firmware hang and:

"Indicated Virtual Disk has transitioned to Stalled Too Long" and TC:0x4238011f "The host port has detected a CSM reset after 60 minutes"

[Scatty]

Similar issue with one of our EVAs today. Tried deleting a Vdisk in a DR group (the only Vdisk in the DR group), which failed. CV then left with a Vdisk that couldn't be removed (saying it's part of a DR group), but no DR group remains! Failed to remove using SSSU as well. Decided to restart controllers in a controlled fashion. Controller one rebooted via WOCP - OK. Status good. Controller two rebooted - status good. Then (immediately), controller one goes into state "unknown" and the EVA stops presenting its Vdisks. All ESX hosts hang. On the phone to HP, India, all afternoon. Now waiting for HP engineer to attend site with his magic serial cable.