increased memory size in VM, now task stuck in progress

[adiloret]

Hi,

I'm using an HP ProLiant DL360 G5 with a vanilla install of ESXi 4 (haven't upgraded to U1 yet). It has 16GB of memory of which only 7GB is used. The server has been up for well over 136 days and running smoothly.

I noticed one of the Ubuntu 64 VMs had began consuming swap space (due to the workload on it), so I decided to give it more memory. I shut the VM down, and then went into the settings (via vSphere) and increased the memory from 512MB to 768MB...a seemingly simple task that I've done several times before.

Upon pressing OK the Task List in vSphere shows "Reconfigure virtual machine" as "In progress"...it has been that way for over two hours now. The task cannot be canceled either.

So my question is, has anyone seen this before and is there a remedy other than rebooting the ESXi server (which I hope works)? I can last another day or two without the VM, but I need to get this resolved one way or another soon. Will the task ever timeout/error?

I went into "unsupported mode" on the ESXi server and ran the following commands that hopefully can help someone debug:

Task list:

~ # vim-cmd vmsvc/get.tasklist 192

(ManagedObjectReference) [

'vim.Task:haTask-192-vim.VirtualMachine.reconfigure-4376'

]

Attempt to cancel:

~ # vim-cmd vimsvc/task_cancel haTask-192-vim.VirtualMachine.reconfigure-4376

(vmodl.fault.NotSupported) {

dynamicType = ,

}

The log messages (the VM name is Crusher1):

Mar 25 16:28:05 Hostd: Task Created : haTask-192-vim.VirtualMachine.reconfigure-4376

Mar 25 16:28:05 Hostd: Reconfigure: (vim.vm.ConfigSpec) { dynamicType = , cpuAllocation = (vim.ResourceAllocationInfo) null, memoryAllocation = (vim.ResourceAllocationInfo) null, cpuAffinity = (vim.vm.AffinityInfo) null

Mar 25 16:28:05 Hostd:

Mar 25 16:28:05 Hostd: State Transition (VM_STATE_OFF -> VM_STATE_RECONFIGURING)

Mar 25 16:28:05 Hostd: ModeMgr::Begin: op = normal, current = normal, count = 10

Mar 25 16:28:05 Hostd: Current VM Tracking state: disabled

Mar 25 16:28:05 Hostd: Setting VM's tracking state to disabled.

Any help is greatly appreciated.

[Callifo]

Have you tried restarting the management agents on the console? This has saved me countless times when tasks have stopped working.

[adiloret]

Thanks Callifo for the tip.

I attempted to "Restart Management Agents" from the Management Console, but it didn't work. I tried the option a few times.

So then I found information on how to do this from the shell (unsupported mode):

• temporarily disable automatic startup/shutdown of VMs from the vSphere client first (otherwise they will shut down with the next command)

• run: /sbin/services.sh restart

- the services will shut down and restart in a few moments

• re-enable automatic startup/shutdown of VMs

That did the trick. It restarted all agent services, while leaving the ESXi 4 server and all VMs still running. The task was no longer running.

Just as an FYI, when I checked the config of the VM that started this mess, the memory allocation was still set to 512MB. So the configuration change either rolled-back or never took place when the task hung. I was able to then successfully change the memory with no issues.

Cheers