XP-VM crashes ESXi 5 after a few minutes - see the PSOD below
any ideas ?
MCE's are hardware problems. Are you running ESXi as a VM? If not, I would check CPU/RAM of the ESXi Host.
Hi Troy
thats not my machine
all I know at the moment is that the ESXi runs on real metal.
Several VMs are running fine but a XP-VM that uses special network monitoring apps crashes the VM.
I provide more data as soon as possible
You can check out the below KB, but more time than not it's usually a bad core or dimm within the host hardware
Now I have the vmware.logs and vmx-files used
please download them from here
http://www.sploing.org/temp/vmware.log.rar
I'm not too good at interpreting the PSOD's. With that said, the PSOD identifies a hardware problem. There should be a zdump file at the /root (assuming you configured a scratch partition). The zdump may tell us more. Other than that, I would start running hardware diagnostics on this server. You may also want to take the zdump and get it to the hardware vendor for them to review.
This sounds like AMD erratum 383 - CPU Core May Machine Check When System Software Changes Page Tables Dynamically.
ESXi 5 should work around this erratum by disabling large pages. Have large pages been specifically enabled for this VM? I can't get to sploing from within the VMware firewall.
Hi Jim
thanks for your interest
I attach the files I have
Yikes! This is a Bobcat CPU. ESXi only applies the erratum 383 workaround for Barcelona CPUs.
Try adding the following configuration option to /etc/vmware/config, as this should apply to *all* VMs:
monitor_control.disable_mmu_largepages = TRUE
thanks a lot
I report as soon as I hear the results from the user
Bobcat is usually a mobile part. Is this system on the HCL?
I seriously doubt that
that guys said he uses AMD Fusion E-350 Mainboard - dont think that that is a server mainboard
I only posted this issue as it looked like several other VMs were running fine and only one special XP-VM produced this PSOD in a reproduceable way
The workaround I suggested only applies to VMs using hardware-assisted MMU virtualization. I see that this problematic VM is using BT (the default for Windows XP). You will also have to change the preferred execution mode of this VM to "AMD-V/RVI" for the workaround to be effective.
Message was edited by: jmattson - Specifically, hardware-assisted *MMU* virtualization.
ok - thanks again
did I ever mention that I dont like customers that go home at 5 in the afternoon while we try to help them ?
Hi guys,
it's my machine that brought to you this nice colourful PSOD message. Thank you very much for your help with this problem, especially to continuum for starting this thread.
The mainboard is not a server board, on the contrary, it's a very slow machine. I devided to use the E350 because it's cheap, energy saving and comes with virtualisation features. It's not the fastest thing but perfect for my small home server.
The problem came up when i installed Win XP Prof. Another VM runs with Win XP Home for month without any problems. In both cases i used the "default" settings and just changed some basic things like ram size, nics... . One of the XP Prof machines crashed the ESXI5 already during Windows installation, that's scary!
jmattson: I added the line to the /etc/vmware/config and switched the VT Option to "VT-X/AMD-V + EPT/AMD RVI". The Machine now runs fine for the last 10 minates, hope it will stay like that I'll tell you the results after some runtime.
continuum, had to go home at half to five today to work on some hardware... my bike needed a new chain
For more information, see http://support.amd.com/us/Processor_TechDocs/41322.pdf. Look for '383.' It comes as a bit of a surprise to me that this bug was not fixed in Bobcat. Unfortunately, since there are no Bobcat systems on the ESX(i) HCL, our Bobcat testing has been rather limited.
Some days are gone since the small modification of the ESXi and the system still runs perfect without crashing. Thank you very much for your help
Regards
Chris