sgadsby
Contributor
Contributor

LGTO_Sync hangs VM -- how to solve?

G'day,

We backup a number of VMs with VCB 1.5 on ESX 3.5. For one VM in particular we are using the mount-method to backup file data incrementally.

Periodically however (1 in 100 vcb mounts?) the VCB script to mount the drive from the SAN fails and the VM becomes completely unresponsive for 15 minutes at which time the VCB script times out.

After those 15 minutes the machine picks up where it left off, no need to reboot.

I suspect this issue may be to do with LGTO_Sync in the VM rather than the snapshot mechanism on the server. Thoughts?

If so, then how can I determine what process on the server is preventing the sync from completing?

The vmware/hostd.log shows the 15min gap:

\[2008-12-08 15:00:26.018 'TaskManager' 22379440 info] Task Created : haTask-16-vim.VirtualMachine.createSnapshot-45399

\[2008-12-08 15:00:26.021 'vm:/vmfs/volumes/47ce2fbb-b0846258-ccff-001a645a922e/TC-FP-01/TC-FP-01.vmx' 22379440 info] State Transition (VM_STATE_ON -> VM_STATE_CREATE_SNAPSHOT)

\[2008-12-08 15:00:26.022 'vm:/vmfs/volumes/47ce2fbb-b0846258-ccff-001a645a922e/TC-FP-01/TC-FP-01.vmx' 22379440 info] Trying quiesced snapshot via backup agent...

\[2008-12-08 15:15:28.157 'vm:/vmfs/volumes/47ce2fbb-b0846258-ccff-001a645a922e/TC-FP-01/TC-FP-01.vmx' 22379440 info] Starting to snapshot VM.

The Windows event log has:

8/12/08 3:00:26 LGTO_Sync Flush Completed.

8/12/08 3:00:26 LGTO_Sync Flush Completed.

8/12/08 3:15:28 LGTO_Sync Sync Stop done.

8/12/08 3:15:28 LGTO_Sync Sync Stop done.

I am aware that I could disable the LGTO_Sync driver, but the cost is crash-consistent backups as opposed to application-consistent, which is not an option for the client.

Your advice appreciated,

Simon.


--

I need to read "Xen and the art of VMware sales" Smiley Happy

-- I need to read "Xen and the art of VMware sales" πŸ™‚
0 Kudos
4 Replies
sgadsby
Contributor
Contributor

Further to this, it seems that the machine is not totally unresponsive during the 15-minute period.

RDP sessions are ok, however all network mappings to the server fail during the period.


--

I need to read "Xen and the art of VMware sales" Smiley Happy

-- I need to read "Xen and the art of VMware sales" πŸ™‚
0 Kudos
apatel1
Enthusiast
Enthusiast

We've tried increasing the Service Console RAM to remedy occasional VM hangs during VCB backups. The default is 272 MB and it can be increased to 800 MB.

Hope that helps! Please help me out by marking my response as "helpful" or "correct" if you feel that it was useful!

-Amit

Please help me out by awarding points for a "helpful" or "correct" response if you feel that it was useful! -Amit
0 Kudos
sgadsby
Contributor
Contributor

Would you recommend setting it to the maximum 800 or something lower?

What metrics can you use to decide?

Thanks for the suggestion.

--

I need to read "Xen and the art of VMware sales" Smiley Happy

-- I need to read "Xen and the art of VMware sales" πŸ™‚
0 Kudos
apatel1
Enthusiast
Enthusiast

I've usually set it all the way to 800 just for piece of mind. A few hundred MBs usually isn't a big expense at the end of the day, but I can understand if memory is scarce. You could look at esxtop on the Service Console (in "m" mode) to see how much of your Service Console memory is being utilized during the backup window.

Hope that helps! Please help me out by marking my response as "helpful" or "correct" if you feel that it was useful!

-Amit

Please help me out by awarding points for a "helpful" or "correct" response if you feel that it was useful! -Amit