HPE StoreVirtual Storage / LeftHand
cancel
Showing results for 
Search instead for 
Did you mean: 

Bizarre VSA Upgrade Behavior - 12.5 to 12.6

Princes
Advisor

Bizarre VSA Upgrade Behavior - 12.5 to 12.6

Hi,

Running the upgrade from CMC 12.6 on a 2-node 12.5 Cluster with FOM.

Platform is a 2-node vSphere 5.5 u3b on DL380 Gen9.

Upgrade proceeds normally until it come time to reboot the first VSA. Instead of rebooting, the VSA just locks up. After waiting ages I have resorted to a manual reboot which brings it back, however a duplicate VSA now appears in the management group and the storage services of the VSAthat is still showing in the cluster no longer start.

The only way I have found to get this back without loosing all the volumes is to set the cluster storage system to repair, remove the duplicate storage system of the same name from the management group, re-add it to the management group and then exchange the storage systems. However this triggers a very long and time consuming restripe.

Anyone ever seen behavious like this before? It is repeatable......!

 

3 REPLIES
Stor_Mort
HPE Pro

Re: Bizarre VSA Upgrade Behavior - 12.5 to 12.6

Firstly, is this a VSA 2014 or a legacy (9.5) VSA? The legacy VSA can be unstable if upgraded past 11.5. Next thing to check, have you allocated enough memory according to the VSA Installation and Configuration Guide (www.hpe.com/info/storevirtualdownloads)?

I am an HPE employee - HPE StoreVirtual Support
Princes
Advisor

Re: Bizarre VSA Upgrade Behavior - 12.5 to 12.6

Hi,

They are VSA 2014's (originally deployed as v12.5 nodes). I have allogated and reserved the memory in accordance with the guide as well.

Since originally posting, I have recreated the entire cluster by deploying fresh v12.6 nodes, however the issue remains. Again, it doesn't happen every time (maybe 1 in 3 node reboots), and seems to particularly affect the node that is not currently the 'LefHand OS Connection' master.

Stor_Mort
HPE Pro

Re: Bizarre VSA Upgrade Behavior - 12.5 to 12.6

Hi Princes,

That is odd. I'm thinking that there may be a communications problem between the VSAs. The simplest configuration is to run all services on one network segment using one virtual NIC. If you have more than one vNIC assigned, they must be on different subnets and the functions must be assigned with the Network > Communications tab (example screen shot attached.)

LeftHand OS interface is used for inter-node communications. iSCSI interface is for host connections. The Management interface (if used) would be the one with an IP gateway so SNMP, SMTP, NTP traffic can be routed to a management network if necessary.

I am an HPE employee - HPE StoreVirtual Support