EVA6000 downgrade from 6220 to 6110?

[agent006]

In October we upgraded our 6000 to 6220 firmware. Since then we've been plagued with storage problems that both HP support and VMWare seem unable to resolve.

Is it possible to downgrade the firmware back to 6110?

[Víctor Cespón · Accepted solution]

No, you'll have to un-initialize the array, losing everything.

The upgrade / downgrade matrices are in this document:

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c01942570/c01942570.pdf

[Stefan Sjogren_1]

Hi,

Could you provide some details about the issues you have been having with 6220 and VMWare?
We are running VMware and have multiple EVA:s running 6110. We recently had one EVA crash due to a known issue in 6110 and are planning upgrades to 6220.

regards,
Stefan

[agent006]

Problem is that storage appears to time out on a host, triggering HA to load up VMs on another host. HP have identified this as caused by vdisks changing controller mastership due to incorrect path balancing on the VMWare end. Even after changing that, and also replacing HBAs the problem still exists. All this never happened with 6110.

HP are working on this latest occurrence now. VMWare are very quick to wash their hands of support requests the moment a storage issue is hinted at.

[Stefan Sjogren_1]

Have you tried to manually balance the paths on the VMWare end?

In ESX3 we had problems with all LUN:s using one path only, choking that EVA controllers host port. This was resolved by manually balance LUN:s across all paths in ESX. In vSphere we use roundrobin.

[agent006]

HP thought it was the manual balancing at fault. We had balanced the paths but each host had a different path set to each lun. So to lun1, for example, host1 would take one path and host2 would take another and host3 would take another.

They thought that the EVA was moving the disks between controllers as traffic came in on different paths. We've now set all hosts to use the same path for the same LUN but the problem remains.

[Stefan Sjogren_1]

Have you checked the EVA:s host ports if they are unbalanced, perhaps caused by non-VMWare hosts?

[agent006]

Host ports look OK. We have no san connected non-vmware hosts.

[Herman Sugeng_1]

Hallo,
If your EVA does not shows problem like describe in Service Advisory c01914371 than is balancing EVA-VMWare probably the solution.
Our experience is with VMWare you have to configured the ownerhip of the VDisk in CVE, remeber which controller it is. Then in VC go to all your VMWare server in the same HA and fixed to the same EVA controller, do this for all you VDisk. Try to balance the VDisk controller and HBA in your VMWare servers. If it is good balance you will not get the messages "chnaging cotroller mastership . . " anymore.
Best Regards,

Best regards,

[Gordon Sjodin]

I have a very simmilar problem and did load ballance all of my paths. I found that if a controller was rebooted or some other event I could loose the proper setting and in fact be flipping to the wrong controller. I have since upgraded vmware and utalized round-robin load ballancing and have not had any probelsm since.