application failure when Vmware removes snapshot

[miklaw2]

I have been trying to get this not to happen... I have tried without the VMware quiesce and with... I have servers that are accessing other servers ..some SQL some MQSeries, and it looks like when the removes the snapshot cause applications to fail... and not all of them....
I was wondering if anyone else has had a similar issue and how they dealt with it. We just bought Veeam and when we were testing didn't backup any of these type of servers. (my bad)
Thanks in advance.

background we are VSPhere Vcenter Server 4 build 208111
All of my hosts are ESX 4.0.0 244038
The servers in question are all Windows 2003 SP2 32bit vms

[Gostev]

Hello, what sort of failure are you talking about, and what is your support case ID?

[miklaw2]

I haven't opened a support case yet... truthfully wasn't sure if it was a Veeam issue or a vmware issue. mostly timeouts...
For instance we have a server that is running SQL server traces ...and 2 of the five failed. We have some processes that run on a server and connect to MQServer Queues on another server... and when snapshot is removed some of them lose connectivity to the MQ server.
I don't see any errors on the sql trace server other then failed in the process\trace window. On the MQ processes I see MQRC_CONNECTION_BROKEN

[Gostev]

Likely VMware issue, as we do not do anything at snapshot removal (except sending API call asking VMware to remove this snapshot).

Could be poorly written SQL app as well, as 2-3 sec VM stun is normal for final commit during snapshot removal... well written application should not have any problems with such timeouts.

Please check if there are existing snapshots (other than ones Veeam creates) on VMs where you see these timeouts, as this is what can be causing timeout issues on snapshot removal with ESX 4.0.

There is also an existing topic about this issue on the forum where other customers exchange some troubleshooting tips and ideas around this problem. Look for it, as it might be helpful.

[rgs3]

I have this issue as well. I get 20-30 second timeouts when Veeam commits the snap. VMware says its veeam, Veeam says its VMware.

[Gostev]

Richard, please contact our support and they will provide step-by-step instruction of reproducing the issue without Veeam Backup installed.

For any 3rd party product, VMware snapshot removal is completely isolated to a single API call, RemoveSnapshot(). Essentially, all 3rd party product can do is ask vSphere to remove certain snapshot. Everything else is handled by VMware vCenter and ESX server code (including VM stun on snapshot commit, which I think what is causes the issue here for you). 3rd party products are not supposed, and moreover simply not provided by any means of affecting any part or step of the (very complex) snapshot removal process.