I have ESX 3.0.1 on a server...
About 5 vm's running on it...
My problem comes in when the power fails and the UPS takes over for 30sec. before the generator kicks in...
We once had power spikes and the ups took over, even then the VM turned off...
One of the five VM's on the ESX server turns off every time...
I do not know why...
any help would be grateful...
Sounds like a problem with your UPS / hardware configuration and not VMware. If the UPS doesn't take over quickly enough then there will be a spike causing havoc on the host. If the spike was long enough, you had a hard reset (unplug/plugin powercord idea), if it wasn't long enough, but still a big spike, it would depend on your PSU(s) if they are good enough to handle the spike. If those are not capable of suppressing the spike then who knows what happens with your VMs.
So i guess that there's not much left to do for you as test... unplug the power while your host is running. Is your UPS quick enough in keeping your host smoothly running?
Tip: Test this when in the bios (or before your host runs VMware) first, so that VMware doesn't get into an inconsistent state.
The UPS is not the problem, it is a big UPS and it can run double the amount of blades we are currently running...
Why is the rest of the VM's still running?
I can understand if all VM's would drop if the UPS wasn't sufficient...
I can tell U that it is not the UPS...
If it is not the UPS then the ESX server wouldn't know that the power dropped unless you already installed a UPS agent on ESX and you did not mention that.
So sorry, but i still fail to see this being a vmware issue.
--
Wil
Do you think it could be a host isolation problem? If the network switches go down the host may think its isolated and by default (isolation response) will start to shutdown VMs, so they can be restarted on other systems. Is all of your switching on the UPS as well?
There are not agents or software running on the server that is talking to the UPS, maybe something left after it was P2V'd from a physical box?
What OS is it? Anything in the logs for the OS or the ESX server in regards to that VM?
Considering awarding points if this is of use
Good tip Rod, why didn't i think of a switch not being on redundant power... ![]()
In that case maybe we should also link to your document:
--
Wil
That specific rack with 2 Servers a SAN, switches and routers is on the UPS.
It is a APC and it has got network capabilities, but that I disabled because I thought is was that what turned the server of over TCP/IP...
Once a gene I say it is not the UPS's fault...
4 of the 5 VM's is still running, why then not all dead?
Okay, so its something in the VM. What do the log files in the guest OS and VC tell you about it?
Does the VM crash or is it a graceful shutdown?
I think you are going to have to do some more research into these areas.
Considering awarding points if this is of use
Thanks Rodos,
It just drops (crach) the VM , VC?
Yes, and I have no idea where to go and start looking for the problem...
Thanks for all your help so far...
No problem.
When you say it crashes the VM, VC which one? Is your VC a VM? Can you be more specific? Which is it the VM or the VC or both?
Its just so weird it has us intrigued as to what it could be.
Have you checked the log files yet, entries close to the time of a shutdown? That is somewhere to start looking?
Really can't see how this can be a VMware bug or issue from what you have described so far.
Keep digging, don't give up. Thats what separates average from great IT people. Who know the root problem may be something that could really cause you grief down the track. No stone unturned. If it does not make sense its not right.
Rodos I found the problem, it was the ups software...
The UPS has got a TCP/IP shutdown fetcher...
But there is a local agent what you can install as swell...
Agent was installed on the server and it running in the background...
I did not know that so I checked the services and it was running...
Like you said, Keep digging, and don't give up. That's what separates average from great IT people. Who know the root problem may be something that could really cause you grief down the track. No stone unturned. If it does not make sense it's not right.
thax
Cool to hear that you found the problem.
Please mark your question as answered and assign helpful/correct buttons for awarding helpful/correct answers from people that helped.
This will help people that search the forums and keeps everyone motivated.
See also here: http://www.vmware.com/communities/content/quicktour.html#c61912
thanks!
Wil
Wonderful!.
So my comment early on
There are not agents or software running on the server that is talking to the UPS, maybe something left after it was P2V'd from a physical box?
Was on the money. I think Wila was onto it too.
Sorry guys my mistake...
LOL, don't be sorry and no mistake.
I remember a story about a teacher who told his students they had to go and talk to the teddy bear before they came and asked him a question. Just by going and explaining the problem to the teddy bear they often solved it themselves. Always loved that story.
