srazaddn
Contributor
Contributor

CVA 2.2 FC SAN certification: whenever storage goes away the windows host on the storage get corrupted

While running the CVA tests for single initiator- if both the paths to the storage go away for a little while then the windows hosts get corrupted. The storage is in Write through mode and I am running continuous verifies. It happens to the VMs that are located on the SAN. Is this normal? Are the any workarounds for this.

Is it ok to have all the VMs locally and have their primary storage on the SAN?

Thanks,

Salman Raza

0 Kudos
2 Replies
RParker
Immortal
Immortal

It happens to the VMs that are located on the SAN. Is this normal? Are the any workarounds for this.

The first question I would ask is why are you losing SAN connectivity if you have redundant paths. If you do, and it still happens you have a major infrastructure problem. Should Windows keep cache of data forever until the drive comes back? Probably not, since that's it primary source of information retrieval and storage. There is some amount of resilience in Windows (depending on what you are doing) we have Windows VM's that go offline for hours (old hardware that fails) but as soon as I restore the connection, Windows get's rebooted (sometimes it comes back), but no corruption.

Is it ok to have all the VMs locally and have their primary storage on the SAN?

If their primary storage is on the SAN, I assume that's the drive that get's corrupted. Why do you have the storage split? You can't migrate them easily doing this. So what part of the local environment is on local disk? Have you tried to isolate why the SAN loses connectivity?

0 Kudos
srazaddn
Contributor
Contributor

Hi Parker,

I am trying to certify an FC storage array. To do that I have to run a bunch of tests that have specific topology requirements set by Vmware. One of the requirements is to have 4 VMs on the SAN, so my second question was related to the test setup. If I can have the VM locally, and the OS on the SAN lun.

Your answer to the first question makes sense. The problem in running the interoperability tests is that there is a high probability of either the ESX server or the SAN not liking each other all the time which right now is resulting in seting up the VM host from scratch.

Thanks for the response,

Salman Raza

0 Kudos