chris_gardner
Contributor
Contributor

VMotion / HA / DRS quandry

Jump to solution

We recently purchased Infrastructure 3 with 2 ESX licenses and 1 VirtualCenter. We got the package with HA and DRS. I've been setting things up and I'm a little confused.

We clustered the 2 ESX nodes. What I'd like (don't know if it's possible): if a node fails, the VM is automatically moved over to the other node in the state it is in.

My understanding:

1.) VMotion will move a VM over in the state it was in, but it's manual.

2.) DRS will move VMs over in the state they were in.

3.) If a node fails, HA will move the VM over. However, it in doing so the VM will be rebooted.

Is this correct?

Also, I'm running into a warning regarding possible errors on hosts with snapshots if I VMotion between nodes. I would think this would be a fairly common concern, but Googling it only results in one forum post with the error message. What is that about, and should I be concerned?

0 Kudos
1 Solution

Accepted Solutions
Rumple
Virtuoso
Virtuoso

Your understanding is completely correct on all 3 points.

There is another thread about vmotion with snapshots and the concensus is that it gives a warning but does still work correctly.

View solution in original post

0 Kudos
2 Replies
Rumple
Virtuoso
Virtuoso

Your understanding is completely correct on all 3 points.

There is another thread about vmotion with snapshots and the concensus is that it gives a warning but does still work correctly.

0 Kudos
chandlm
Expert
Expert

We clustered the 2 ESX nodes. What I'd like (don't

know if it's possible): if a node fails, the VM is

automatically moved over to the other node in the

state it is in.

My understanding:

1.) VMotion will move a VM over in the state it was

in, but it's manual.

If a host fails, VMotion will not be able to move the guest really since it is powered off at the time of the failure. VMotion is for moving guests from host to host as live servers. Correct in that it is manual.

2.) DRS will move VMs over in the state they were

in.

See #1, DRS is basically automated VMotion with logic behind it. It decides that Host A is overutilized while Host B is underused so it will move any VMs configured to move from A -> B

3.) If a node fails, HA will move the VM over.

However, it in doing so the VM will be rebooted.

Yes, you need to keep in mind that if a host becomes unavailable it would no longer be running the VMs (unless there is a lapse in heartbeat communications but that's another complete discussion). That VM could be automatically restarted on another member of the HA Cluster.

Is this correct?

Also, I'm running into a warning regarding possible

errors on hosts with snapshots if I VMotion between

nodes. I would think this would be a fairly common

concern, but Googling it only results in one forum

post with the error message. What is that about, and

should I be concerned?

Not sure I can help on this one. Do they move OK? I believe that VMotion wants to move VMs without redo logs, etc... so that's probably why. Does the VM move successfully and could you commit any changes that are being saved due to snapshots?

Does that help at all?

Mark