Vmotion Times Out But Completes Anyway

[sheetsb]

I am running ESX 3.0.1 enterprise with what I believe are all the current patches. VC is 2.0.1 build 40644. I have created a test HA/DRS cluster and wanted to migrate a few VMs to see how the cluster performs. On this particular cluster I migrated one VM without problems so I started to VMotion a second. The migration proceeded to 90% and then stopped for an extended period. After a while I received the process timed out message. However, the VM appeared in the cluster on the new host.

Anyone seen this before or have any suggestions as to why I got the error, even though the migration completed? This isn't very comforting for my first testing of HA/DRS.

Thanks,

Bill S.

[waynegrow]

I have seen something similar a couple of times. I believe the timeout is just reporting it back to VC, and not necessarily a ESX/VM problem.

[geekatcmu]

I recently saw behavior similar to this in my VC console. Shortly after this started, I ran out of space in my VC database (doh!). Adding more tablespace (and setting it to autoextend) seems to have rectified the problem.

[masaki]

Be careful.

When I saw this behaviour in the past there could be some side effect like file duplication.

[sheetsb]

What sort of file duplication? I'm not relocating the VMs so what is being duplicated?

Bill S.

[Texiwill]

Hello,

To determine the problem you can look at the VC logs.... C:\Program Files\VMware VirtualCenter 2.0\tomcat\logs as well as look at the /var/log/vmware/hostd.log on both the source and target hosts.

This will help you to narrow down the problem.

Best regards,

Edward

[masaki]

You're right the problem I was talking about it's about relocating (cold migration) and not vmotion.

Another idea. do you have the vmkernel port group on a dedicated vswitch and physical nic/s ?

If not try separating vmkernel port group.

Try to post here your vswitch config:

esxcfg-vswitch -l

[sheetsb]

Each ESX host has three vswitches one for kernel, one for VMs and one for Vmotion. The kernel and vmotion each have a separate, dedicated 1gb NIC. VMs have two.

Bill S.

[Texiwill]

Hello,

You will need to review the logfiles or post questionable parts of them for us to determine the problem. It could be an issue unrelated to the network.

Best regards,

Edward

[masaki]

Increase the VC log level.

Look or post logs.

Revert VC log to normal.