Infrastructure Software

Hello all,

I’ve been experiencing an issue with our Oracle VM backup process, this issue is localized to a 10g Oracle Enterprise Application Server running as a paravirtualized VM within the environment. The VM infrastructure contains two Dell PowerEdge servers running Oracle VM 2.2.2, the filesystem is clustered with ocfs2 and is shared out with a Dell MD3000i via iSCSI. The backup process is relatively simple and works as follows:

1. Suspend all running machines in the cluster using ovm cli
2. Create a snapshot of the /OVS partition using the Dell MD3000i
3. Resume all suspended machines via ovm cli
4. Mount and backup the snapshot on a separate system

After creating the snapshot and resuming all the suspended virtual machines (about 10 VMs total), all VMs except the 10g Enterprise Application Server VM resume in the state prior to issuing the vm suspend command. I did a bit of troubleshooting only with this VM, and if I suspend it while all the Oracle services are running and then resume it (exactly the same as when the backup is run) the previously running services are all offline. On the infrastructure side, the listener is down, the database is idle, and there are no opmn services running. On the apps side all the opmn processes are down as well.

If I pause and un-pause this VM, everything comes back up fine. This issue only seems to happen when suspending and resuming the VM to halt IO before creating a snapshot. From what I remember when configuring the backup, the VMs need to be in the suspended state in order to create a usable snapshot. I tried looking through the VM documentation again to determine the differences between the suspended and paused states but the information was very generic (I may have been looking in the wrong place).

Does anyone have any clues as to why the Enterprise Application Server is halting Oracle services when it’s suspended and resumed? On a separate note, does anyone know if I can pause a VM instead of suspending it to create a viable backup of it with a filesystem snapshot?

I appreciate any responses and can help with any additional questions if needed.

Thanks!

Edited by: npawelek on Jan 10, 2012 10:43 AM