Virtual machine CPU usage spikes and remains abnormally high after VMotion in a VMware DRS enabled cluster, IS BACK!

[Speedbmp]

Ok I have followed this and it does confirm that I have the problem however I

am now running 3.5.0 84374 post update 1.

I am still having this issue, i have

tried what is said in the KB and nothing seems to take care of the problem, if I

turn off DRS the problem does go away.

Stephen

[Speedbmp]

Bump,

Any Idea's?

Stephen

[admin]

Are you using custom or default CPU share values for the VM's?

Try experimenting with increasing and decreasing the shares to see if things improve or worsen.

[dodell]

I had this problem in the following article, but applying update 1 fixed the problem: . It was working well for a couple weeks.

I just installed the patches that were released on May 1st and the DRS problem came back. 1 minute after I disable DRS everything is back to normal. I browse to the advanced setting of Mem.VMOverheadGrowthLimit and it is set at -1. Update 1 fixed this problem by making this setting a 5. The patches I applied removed the DRS fix that was applied in update 1.

Message was edited by: Ken.Cline to shorten the URL

[Speedbmp]

Yes i think you are right the problem is back, it was fixed but with the latest patches it seems to be back.

Stephen

[Speedbmp]

Bump,

Does anybody else have this problem? the problem WAS fixed but now it's back with the latest patches.

Stephen

[JoshuaP]

It's happening here too.

[JoshuaP]

I just noticed http://kb.vmware.com/kb/1003638 says "Note: This issue is resolved as of VirtualCenter 2.5.0 Update 1". My ESX Servers are up to date, but my VC server is not.

[Speedbmp]

I have already updated to update 1 for VC and ESX, this took care of the problem. but when the latest patches came out it broke it again.

Stephen

[Speedbmp]

Bump,

How many people are having this problem after 4-30-08 patches? the problem was fixed when update 1 came out. then when i installed the 4-30-08 patches it broke it.

Stephen

[JaniT]

Speedbmp:

We were also experiencing the same problem on ESX cluster with the latest updates installed.

I found this discussion after spending a morning trying to figure out what on earth was going on - all VM's were performing extremely poorly due to high CPU usage after migrations. After reading this thread I disabled DRS and in a few minutes everything was running smoothly again! Later on I re-enabled DRS and haven't had any problems in a couple of days (I left DRS set to "partially automated" so no migrations so far until this is resolved for good).

Jani

[John_S1]

Same problem here. Not all machines exhibit the behavior. All are using default settings, nothing custom. Fixed before patches, broken after patches.

[Rumple]

Someone should teach vmware how to use snapshots as it seems they are constantly reverting their code after fixing something.

Either that or to use a code repository so they stop keeping multiple different copies of the code...

[Speedbmp]

So whats the best way to get this fixed? call VMware or wait tell they issue another a fix next month?

Stephen

[frankagricola]

the trick that solved the problem for me was:

turning off HA and DRS on cluster

create new cluster

move ESX hosts to new cluster

move vm's to new cluster

turn on HA and DRS on new cluster

[John_S1]

We opened a ticket with VMware and the end answer was they know the problem exists and it will be fixed in Update 2. No timeframe on the release date though.

[Speedbmp]

I am glad they are aware of the problem. for right now i will just have to keep DRS disabled.

Stephen

[MalcO]

Would be useful if there was a method to uninstall vmware host patches.

[Speedbmp]

True, it would also be nice to know which of the patches broke it.

Stephen

[MalcO]

Indeed, though would be possible to find out if patches could be removed (albeit somewhat tedius).