9.5 Upgrade Experiences?

[Paul Hutchings]

I'm running 2 clusters, one 2 node P4500 cluster and one 4 node P4300 cluster.

Both are multi-site with a single VIP, so basically a stretch cluster (primarily due to lack of routable iSCSI in vSphere MPIO).

I've not yet upgraded to 9.5 as I want to give it a little while for the dust to settle and any bugs or issues to emerge, plug I didn't see any killer "must have now" features over 9.0.

I know there's a thread that mentions 9.5 features and a few people have mentioned that they've upgraded, but I thought a dedicated upgrade thread might be worthwhile.

So far I've only upgraded my CMC.  First impressions, does the job but that SAN summary screen is a little strange IMO.  Perhaps it makes sense if you have a lot of management groups but to me it just looks cluttered and a little "busy".

[Kyle Thering]

Just wondering if you ever ended up upgrading your clusters to 9.5, I have the exact setup as you have and was just wondering if you ran into any issues.

Thanks

Kyle

[oikjn]

I upgraded without an issue.

pure VSA setup.

two managment groups, setup for remote snapshots.  Remote site is 2-VSAs+FOM, Primary site is also a multi-site cluster with 4-VSAs+FOM in nRAID10. 

Upgrade went without issue and no loss of service from attached hyper-v cluster.  I haven't noticed any significant differences going 9.0 to 9.5 other than the CMC program.

Only "bug" I think I found is that I keep getting messages from one VSA in the remote group that the other's latency is like 2428047324234ms and then get a quick reply that its latency is OK.  I haven't seen any latency issues, but I get flooded with this email  (~50/day) usually under periods of higher load.  I deleted the problem VSA and recreated it, but that didn't change the problem ("problem" being the emails... I haven't noticed any latency issues with the VSA itself). I also moved it to another host and that didn't help.

[WaynehGRC]

I think we have a performance issue since upgrading to 9.5. Same test performed both before and after using IOMeter. Details are:

9.0

IOps 1986.115817

Read IOps 398.425901

Write IOps 1587.689916

MBps 7.758265

Read MBps 1.556351

Write MBps 6.201914

9.5

IOps 377.41583

Read IOps 76.787947

Write IOps 300.627883

MBps 1.474281

Read MBps 0.299953

Write MBps 1.174328

[JohnMurrayUK]

Ouch!  Any other changes Wayne, switches, bond type, volume protection method?

Any idea what the CMC reports your Queue Depth Total as when the IOMeter test is running?

I've managed to get two test P4500G2 shelfs on my desk today (bit loud). One with SANiQ 9.0 and one with SANiQ 9.5. I'll let you know what differences I observe.

[Patrick Terlisten]

Hello,

I did an upgrade in a two Node P4500 G2 Cluster with FOM. Attached was a vSphere 4.1 Cluster with two nodes. The FOM was running as a VM inside the VMware Cluster. During the upgrade the storage services went down. The FOM was first upgraded, then the second P4000 node. After the upgrade of the storage node, services went down. It seems to me, that the FOM wasn't back online as the upgrade of the storage box started.

Does anyone noticed the same?

Regards,

Patrick

[mozturk]

we have 4 nodes p4300 g2 sas storage and one cluster. i upgraded from 9.0 to 9.5 but  performance after the upgrade is extremely decreased .how can i solve this issue. how can i downgrade to previous saniq version.

Pls help.

[5y53ng]

I have a 6 node cluster of P4000 VSAs and I have seen a performance decrease with 9.5 using IOmeter during writes.

[Paul Hutchings]

Have any of you logged and had HP look into the issue?

I'm still on 9.0 as I thought I'd give 9.5 plenty of time before jumping in.

[5y53ng]

Hi Paul,

I'm actually going to give them a call today and see what their take is. Specifically, I only saw a decrease in performance on sequential 64KB writes in IOmeter. I'm using ESX 4.1.