Windows 200x VM clocks running fast

[qsnow]

I've done various searching and there are a various posts on clocks running fast, but a lot of them seem to talk around the guest VMs being a form of linux 2.6.

I currently have things configured as follows:

2 ESX 3.0.1 servers - both SAN connected and configured with DRS/HA

Both ESX servers were configured following the 1339 KB article (and 1318 I think) to configure the ESX servers to point to the stratum time sources.

I then configured the Windows VMs by disabling the 'Windows Time' Service and then enabling the VMWare tools time sync option. If the clock is behind, things catch up successfully and we dont see any problems.

What seems to be happening tho, is that 4 or 5 of my current 11 VMs gain time to fast. I am seeing a 5 to 10 minute time gain in 12 hours or so. Could someone explain to me what is going on/why this is happening and suggest some possible fixes? This is a rather ugly issue, especially for one of my VMs that is running SQL server. Really messes up the logs and DB operations.

[acr]

Have you stopped the Time Service in the VM's

[qsnow]

Absolutely. On all the VMs I disabled the 'Windows Time' service and stopped them. I have also tripple checked them to make sure

[admin]

I've seen this issue with some of my Windows 2003 VMs, but strangely not all of them. The trouble is once the time starts slipping forwards VMTools will not adjust it as it will only adjust the clock forwards, not backwards, and as you say it slips forwards fast!

Personally I think this is a 3.0.1 issue/bug. I gave up trying to fix the issue and set up the Windows Time Service to sync with an NTP source, this has kept the time stable on the problem VMs, but in my mind it's a workaround not a fix.

Sorry I can't be more helpful.....

[qsnow]

FWIW: We are also running 3.0.1. I didn't have any of the clock stuff setup at all in 3.0 so I can't verify if I had the problem in 3.0 or not.

Did you have all the time stuff setup in 3.0 and then it started slipping forward after you upgraded to 3.0.1? If so it sure would point to a 3.0.1 issue in my mind.

How often do you have your windows boxes (running ntp) checking in to get a time update from the windows time server currently? Also did you just point it to the same sources as the ESX server itself (the pool.ntp.org - or whatever it was called)?

[qsnow]

na

[gogogo5]

Do you have Group Policies setup that might reset the Windows Time Service back to automatic and start it?

[zemotard]

Have you check the time synchronisation in the VM Tools ???

[qsnow]

Well, things had been running fine for a while. I was looking at my VMs today and noticed that about 6 out of my 25 VMs had the clock about 4 hours ahead of the current time.

We did have a power failure issue yesterday where everything dropped hard/dirty. The power was out for ~3 hours. Any chance of that causing the VMs to freak out on the clock?

if not, has anyone further seen this problem that might have some additional insight?

[qsnow]

Marked as not answered.