vMotion Errors w/NFS Datastores (error caused by file /datastore/vm/vm.vmdk)

[KCH2]

Clearly this is some kind of networking issue but I cannot figure out what I am missing.

The symptom is the error message above and it happens under two scenarios:

1. Performing a storage vmotion of a virtual machine that is powered off. If you power on the vm it will migrate as expected.

2. Performing a storage vmotion of a virtual machine that is powered on that has vmware snapshots associated with it. If you delete the snapshots the vm will migrate as expected.

I read something about cold storage vmotion data using the management interface to perform a file copy, though I don't know why it would have an issue with my management interface.

Each hosts configuration is 3 vmkernel adapters all sharing a pair of 10Gb ports configured with enhanced LACP using a nexus 5k.

vmk0 - is management and fault tolerance (though I'm not actively using FT)

vmk1 - is NFS (the NFS network is a non-routed network and the storage for that network is supplied by a netapp cdot cluster with an export policy only allowing access from this subnet)

vmk2 - is vMotion

The only think i saw in the logs when attempting vmotion was an error about the mac being changed to 00:00:00:00 and a policy blocking it from happening, but a search on that shows it's a temporary condition that happens during vmotion and is expected.

Any thoughts on why this is happening and how to resolve it?

[mprazeres183]

Hi KCH2,

The situation you are describing is quite difficult to reproduce, I just tryed to do it on my LAB and I saw something similiar but I can't really say to 100% that this is the cause.
So I will give you workaround you can try to do to see if it will solve your issue:

1. First get yourself as many IP Adresses on the vLAN where you have the service vMotion enabled for each Host. If you do not use any vLAN just take the IP Adress from the segment where the vMotion is enabled.
2. Go to Host, Configure, VMkernel adapters and Add a new Adaptor, Select VMkernel Network Adapter, Select an existing Network, Select vMotion and add a new IP Address, make sure to HIT override default gateway for this adapter if the Network is different than the Management Network.
3. Do that on all other hosts.
4. On all hosts select now the OLD vmkernel where you had vMotion enabled and DE-SELECT vMotion as a service. It will turn of that vMotion IP Adress.
5. Reboot all the Hosts by going trough Maintenance Mode, have all the DRS moving the vGuests (This is a test to see if it works and at the same time to reboot all services on the Host)

Now regarding the issues you had with:

1. Performing a storage vmotion of a virtual machine that is powered off. If you power on the vm it will migrate as expected.

Can you send me the error message you get when you try to do so:

2. Performing a storage vmotion of a virtual machine that is powered on that has vmware snapshots associated with it. If you delete the snapshots the vm will migrate as expected.

This can be related to the Snapshot, that could have a .lock, .lck file, in this case a migration would not work.
Anyway please send me the error message.

Best regards,

Marco