Well apart from the Guest OS...
Did you check esxtop stats?
I’m facing the same problem right now
how did you manage to solve yours?
I’ve searched on other forums but I wasn’t able to find a solution
moderator edit: removed irrelevant hyperlinks
You might want to check out KB 2147608 VMware Knowledge Base
6.5 was the last release of hypervisor binary translation mode, and you need to change the virtual machine to be configured as uniprocessor. I wonder if this is the cause in your case?
To answer Oziejfjdjfb, I have not solved my problem yet. My fix was to migrate my VM back to a 6.5 host and pause my upgrade.
To TotesHagopes, I stumbled across that article as well in my later searching over the weekend. I think you're correct, and the removal of that feature is causing the issue. My guess is 6.5 and earlier just "threw away" the halt messages with the binary translation so we didn't get the high CPU utilization. And now that we're no longer doing that, we see the described behavior.
Unfortunately, the article suggests that upgrading to Windows 2000 SP3 or SP4 would fix the issue. That doesn't appear to be the case, as my VM is already running SP4.
For this particular VM, I think I would be able to downgrade to a single vCPU from a resource perspective, but that appears to not be supported (see the first article I referenced):
Note: Downgrading the HAL from multiprocessor to uniprocessor is not supported by Microsoft in Windows Server 2000 and it is not supported by VMware. For more information, see Cannot Select MPS Uniprocessor or ACPI Uniprocessor HAL
Of course, running Windows 2000 here isn't supported either, so not sure this is really an issue. I may try it to see if this is a fix and let you all know. Thanks for taking the time to take a look...
After a little more digging and taking a clone of my VM to do some testing, I found that my issue appears to be tied to an individual processor type for me.
My Windows 2000 VM that was having problems sits on one of my older clusters running AMD processors. When I created my clone, I put it on one of my newer Intel-based clusters and couldn't reproduce the problem. So I moved the original VM to the Intel Cluster and the problem went away. Not 100% sure why, but I'm still guessing that the removal of Binary Translation gave me some unexpected behavior changes. Either way, it's working, so I'm going to leave it as is and still try to get this Windows 2000 VM off the wire. I've not tried across more processor types, but this is what's working for me:
AMD Cluster with the problem : AMD Opteron 6380
Intel Cluster working : Intel Xeon E5-2690 v4
Hope this maybe helps someone out there with the same issue...
We have a couple of major SAP system still running on Windows 2000. After the upgrade to vSphere 6.7 we start to have the problem report.
After analysis, our servers has the latest VMware Tools whch are version 10.0.0.12, latest SP4 + Rollup installed. However, we do have configured them with 8 vCPU with 1 cores each that give 8 sockets total. The KB said that uniprocessor is now required for legacy solution.
We only change our setting for 1 vCPU with 8 cores in our settingsand all servers back to normal.