VM guest issues after moving VMs into resource pools

[Jeffrey79]

My apologies if this has been talked about before. I couldn't find it after an hour of searching, but I have a feeling the wording people use might vary. So anyways, here's my issue.

Freshly installed VC2.5. Several ESX 3.0.2 server all with local attached storage.

I created 3 resource pools, with CPU and Memory Share settings set to Low, Normal, and High respectively. No reservation or limits set. Of my VMs, only a select few have either Low, Normal, or High Shares set on the individual VMs. No reservations or limits set on the individual VMs.

Moving several of my VMs into the 3 resource pools caused issues in some of the VMs. For example, everyone connected to network printers on the VM print server lost their connections and did not get them back until the spooler service on the print server was restarted. Then later in the day our DNS/DHCP VMs starting crashing with "memory errors" saying they ran out of memory. I haven't seen these errors yet. I heard second hand from another admin who I still need to talk further with. Apparently several other applications running from VMs also experienced some sort of issue. I'll find out more tomorrow I'm sure.

So my question is, has anyone seen anything like this before? Are there any known issues with moving VMs into resource pools while the VM is running?

Oh, not sure if it's relevant, but the 3 resource pools are under a DRS enabled cluster. However, like the I said, the ESX server are all on local storage at this point so there is no vmotion migrations going on.

Thanks very much for any assistance at all.

Jeff

[Rubeck]

DRS enabled cluster, but no VMotion due to local storage.... Hmmm.. Why?

Sorry for asking, but I do not see the point of enabling DRS in a cluster with no shared storage...

The only thing I could see that would be good for is the initial placement of new VMs.... althoug I would rather choose the best host candidate manually, IMO.

/Rubeck

[weinstein5]

There should be no issues moving a running VM into or out of a resource pool - as Rubek indicated with your only local sotrage DRS will not work - DRS uses vmotion and vmotion requires shared storage.

Now to you problem which ESX hosts are your VMs running on? do you have them balance across the hosts? How many vms are you running on each host? How much memory is on eacg ESX hosts? How much memory is assigned to each vm?

[Jeffrey79]

DRS is enabled on the cluster in preperation for migration to a SAN beginning next month. DRS settings are set to manual.

I have 6 AMD hosts, various levels of memory ranging from 12gb to 32gb. Each host run on average 10 VMs, but some slightly more or less. None of the ESX hosts are running with more than 80% of their physical memory in use. Most are closer to 60% in use. The VMs range in memory from 256mb to 4gb. Most around 1gb. Almost all Win2k3.

The two DNS/DHCP servers that kicked up the out of memory issues in their System logs both have 512mb of memory assigned, Win2k3sp2. They are running on different ESX hosts. When I try to bring up performance data for either of these VMs I get a message saying, "Performance data is currently not available for this entity". Other VMs performance data seems to be fine.

Thanks for the help guys.