SAN Volumes go down during upgrade to SANiQ 9

[Michiel Siebum]

Hi,

Last Friday we upgraded our HP Lefthand P4500 Multi-Site SAN from SANiQ 8.5 to SANiQ 9.
During the upgrade which was (as instructed)performed by CMC version 9.0, we noticed that the SAN volumes went down.
Not all at the same time, but still...Not what we expected during an upgrade.

Does anyone know about a way to prevent this in future upgrades?

[Michiel Siebum]

I'd like to add that that the SAN cluster is configured with only RAID 10 volumes. So volumes should stay online during the upgrade according to the documentation

[Marcel-p4500]

Question; was your failover manager active?

[Marcel-p4500]

never mind, multi-site should have muliti-managers...

[teledata]

So I'm guessing you had a configuration something like this:

Site A:
Node1 (Mgr)
Node3 (Mgr)
FOM Mgr

Site B:
Node2 (Mgr)
Node4 (Mgr)

I'm assuming it performed the upgrade on one node at a time, and you lost quorum?, or did you maintain quorum, but volumes went offline?

[Michiel Siebum]

Hi, this is our configuration:

Site A:
Node1 (Mgr)
Node3 (Mgr)
Node5
Node7
FOM (Mgr)

Site B:
Node2 (Mgr)
Node4 (Mgr)
Node6
Node8

Quorum should have been maintained as the FOM was available. However... As the CMC was busy with the upgrade I could not see if a Site went down and afterwards there was nothing reported about a volume/site that was down.
We only noticed this within our VMware environment.

I hope this makes the picture more clear...

[Marcel-p4500]

Same here, all volumes are down... i started the upgrade from the CMC and the volumes went down.

[Marcel-p4500]

CHECK YOUR LICENSES... at my site it looks like it has to do something with the MAC adress and license...

beacause there is no license in one of the nodes i think that is why i dont see any iscsi connection tot he san...

but again; i am doing my own support here since HP does not call me...

if anyone can provide more details, yes please

[Marcel-p4500]

update: problems are solved.
What happend was that the mac adress of one node changed in the nic-bond so the key was not valid anymore and the esx hosts could make connections to the san cluster...

With support created a new key and all was backup up again after some scary moments...

[Michiel Siebum]

Hi Marcel,

Thanks for the hint, but it the cause of your inactive volumes must be different then mine.
I checked the licences, but the licences on all 8 nodes are still OK.

Michiel