After the application of 3.5.x update 1 (82663), and VC 2.5.0 update 1 (84767), I seemingly have had almost no end of HA and DRS errors.
My last, and most frustrating one is:
"Unable to apply DRS resource settings on host (Reason: Insufficient CPU resources.).This can significantly reduce the effectiveness of DRS."
I have removed the hosts from the cluster, and re-added them, and it seemed that was solvign the problem, but then, one host that i had done this to has this re-appear AGAIN!
Any thoughts anyone?
Same behavior here on a couple of my ESX hosts. I started seeing the behavior immediately after applying VC 2.5 update 1, both before and after upgrading the ESX hosts to 3.5 update 1.
Strangely it only affects hosts across slower network links from the VC server, seemingly.
Check the CPU reservations for all of your VMs, I had this problem and found a VM that had like a 22Ghz reservation for some reason. Killing that reservation returned the cluster to a normal state.
This issue has resulted in a escalation to engineering to reslove a bug.
Interesting...anyone know if it is reservation/limit settings on some virtuals that causes it?
I noticed a service mgmt-vmware restart will fix the issue for a few hours at least. I also did have some unexplainable CPU limits (in my case up around 40GHz) which I removed and things are holding steady for now.
The workaround for this is to remove any reservation that you might have on the main resource pools.
I have two, PRODUCTION and DEVELOPMENT.
Production has 2/3rd shars of the resources and Development has 1/3rd.
I have HIGH, NORMAL and LOW groups under each root pool.
I set my reservations on the two roots only and let the others fall where they may by default.
It was by sliding the reservations all the way to "none" that resolved the issue temporarilly.
restarting the service you mentioned only causes a re-calc to occur which takes time, and only removes the issue for a while.
I have the same problem here to. A root resource pool with pools under the root pool with, high, normal and low prio shares. On one host i receive that same message:
Unable to apply DRS resourcesettings on host 'esx01' in cluster1 (reason:Insufficient CPU resources.) This can significantly reduce the of DRS effectiveness
it is a 3 host cluster.
someone has a solution?
resolved: Had a vm with a reservation of 5000 mhz instead of 500 mhz
Thread moved to the VI: ESX 3.5 forum
Tom Howarth
VMware Communities User Moderator
