Hi all,
We have a Windows 2000 Server SP 4 running IIS. After a couple of days into production, the machine hangs without further information.
We are able to log-in with Virtual Center, minimize some windows, but that is pretty much it. We can not aces the logs or run any application until we restart the machine. After we restart the machine, it works just fine until this happen again.
We checked the system logs in order to see something unusual, but the only thing we encounter is a event id 6008 aka Unexpected shutdown. Moreover, in the virtual center's events I only see machines being rebooted due to this behavior.
We clone this virtual machine and we have the same behavior in the clones as expected.
Any ideas on what could be happening?
The behavior is consistent of a memory leak, especially if you can ping it when it is hung. You can try un-installing VMware tools for a while and see if the problem still happens, maybe the balloon driver is acting flaky and consuming all the memory on the server over time.
Possibly a memory leak inside the OS or any applications running on it. Have you tried using the VC performance monitor to check out cpu/memory/disk/network. My guess is that this is not being caused by the server being virtual and you would probably have the same issue on a physical server. Is the VM pingable when it hangs? I'd start by monitoring the server physical/virtual memory, you might try increasing the memory allocated to the server. Is your ESX server constrained at all? Any memory/cpu limits set on the VM?
Was the VM P2V'd from a physical machine?
Do you have a backup of the VM from before it started doing this, that you can revert to? (I presume not, otherwise you would have done this!).
Try disabling any unneeded services or programs that run at startup as it could be one of those that's causing the problem.
Chris
First of all, thanks esiebert7625 and Chris_S_UK for your answers.
I have pretty much the same OS and Application configuration in four physical machine with no problem.
When the virtual machine hangs, it is not possible to ping it.
It had the VM with 756 M of ram. I just increase it to 1024 and the other VM clones with 756.
VM Configutation:
CPU:
Reservation: 0 | Limit: Unlimited | Shares: Normal | Shares Value: 1000 | % Shares: 25 | Type: n\a
Memory:
Reservation: 0 | Limit: Unlimited | Shares: NOrmal | Shares Value: 7560 | %Shares: 37 | Type: N\a
Esiebert7625 what should I look for while monitoring the psychical /virtual memory?
Chris_S_UK the machine was install from scratch in ESX. Moreover, we do not have something to go back to. The only suspicious services that could be causing this is Norton and Black ICE. Have you experience some behavior with this?
Thank you all for your help...
Basically load up task manager and look for any processes that are using alot of RAM and steadily growing. Make sure you turn on the Virtual Memory column. Typically with a memory leak the amount of RAM a process uses will steadily grow until it has consumed all the memory on the server and the server becomes non-responsive. Also in VirtualCenter you can look at the Memory performance counters for the VM and look for the same behavior.
First of all, Thank for your response esiebert7625...
I was mistaken in my first post when I stated that the machine did not responded to ping while hang. The machine does responde to pings.
While hang, the machine is using about 5% CPU and the memory:
Memorry Active: KB 42560
Memory Granted: KB 651764
Memory Usage: Percent 8.24
The behavior is consistent of a memory leak, especially if you can ping it when it is hung. You can try un-installing VMware tools for a while and see if the problem still happens, maybe the balloon driver is acting flaky and consuming all the memory on the server over time.
Esiebert7625 you are a Virtuoso. Thank you for you quick response. I will uninstall VMware-tools and monitor the behavior of the VM.
No problem, fyi...don't forget to award points for helpful answers using the Helpful/Correct buttons.
Norton antivirus 10.2 was hanging the servers due to a memory leak... We just unistalled it and went back to version 8.1