prevent failure when coming from VSA version before 10.5 (P4000)

[Björn Seeger]

Hi,

today I would like to share my latest nightmare with vsa to give you a chance to prevent it.

We have two two-site-vsa-clusters with 2 and 4 nodes each. They where installed 2011 with version 9.5 and afterwards upgraded up to 12.5. At two weeks ago we wanted to extend a lun. After opening the CMC all volumes where inaccessible. The vsa's seems to crashed. After reboot it looked ok until we do an iSCSI rescan in ESXi. All VSA's crashed again. The CMC shows many errors like cim down, raid off... With the help of HP we were able to exchange the damaged VSA's and rebuild the data. This cost us 4 days and nights of work and several years of lifetime.

So what was the cause? After that I've analysed the filesystems of the broken vsa's. The root volume was full and had no bytes free anymore. This seems to happen on all VSA's at the same time.

VSA Version prior 10.5 was deployed with a disk size of 8GB. From 10.5 upward the disk size is 32GB per default. Versions prior 10.5 are called Model P4000. The support of this Model will end up 2017!

Admins: if you've got this Model then please have a look at root volume free space in your VSA. And exchange them ASAP! HP support knows the right procedure.

best regards Björn

[Stor_Mort]

Can confirm. The legacy VSA gets increasingly unstable if upgraded past 10.5 because of the small system disk. Move to the supported HP StoreVirtual VSA 2014 as soon as possible.