Some ESX4 configuration is lost after system reboot

[CameroonVM]

Hi Folks,

I've been using ESX4 server for months now .

Lately , after creating some Resource Pools (RP) and adding VMs into these pool , I noticed that after rebooting my system , I lost all the Resource Pools I created.

This configuration loss issue is not only limited to RPs , as I lost also some of the portgroup/network configurations upon rebooting the system.

This problem happened both when I entered "Maintenance mode" and then subsequently rebooted the system ; and also when i typed the "reboot" command in the bash shell (I typed the "sync" command many times before typing "reboot" .)

This problem is becoming a kind of annoyance since I have to recreate the lost configuration every time I reboot.

Is there a way of manually forcing the ESX4 to save the configuration so that I do not keep losing some of my configurations?

Could someone please help ?

Thanks in advance .

Danny

[geddam]

Check hardware, sometimes due to board issue...PCI order changes and you may see some POST errors...which may lead, to loose configuration of host after a reboot.

What version of ESX? What is the hardware? Did it ever worked before?

Thanks,,

Ramesh. Geddam,

VCP 3&4, MCTS(Hyper-V), SNIA SCP.

Please award points, if helpful

http://communities.vmware.com/people/geddam?view=documents

http://communities.vmware.com/blogs/rameshgeddam

[a2alpha]

I have seen this happen on a customer site that config changes aren't saved. It was down to the root partition being full. Might be worth checking this as it isn't shown through the vi-client either direct connection or through vcenter.

On the console of your esx host or through putty type:

df -h

Have a look at the row with a / on it and look at usage. If it is at 100% then that is likely the problem,

Hope this helps.

Dan

[CameroonVM]

Thanks for the help , Dan.

There are 2 ESX4 servers ; let's call them esx1 and esx2 .

I checked my systems and noticed that the root partition("/" ) of esx1 100% used up.

But that of esx2 is 36% used up .

Strangely ,though, both servers still manifest the problem.

Do you have any suggestions as to how to recover/add more space to the root partition of esx1 ?

What could be causing this problem on esx2 ?

Thanks , again

Danny

[a2alpha]

If your two hosts are in a cluster together then the resource pool is using resources from both, I'm going to guess, that if the configuration doesn't apply to one of them then it could well be that it can't keep the config. My first job would be to free up some space on /.

I don't have any quick ways of scanning for the files, like treepie or scanner for windows maybe someone the forums can suggest something. The default / partition size is normally more than enough just for esx so something else I would imagine was added. It may be worth just going through the folders and running ls -lah on them to see if any obvious tars or gzs are lurking. Users home folders and /root would be the places to look.

To resize I would reinstall esx and when the disk setup is doen, use advanced mode and assign it some more space there. You can use something like gparted to resize an existing partition but I have always found it easier to reinstall. If you have vms stored locally then gparted maybe the only option.

Let me know how you get on, if the freeing up of esx1 sorts it, and if you find a tool to search for large files let me know.

Good luck,

Dan

[CameroonVM]

Thanks again , Dan.

I just freed more than 3G of space from esx1 and now I have 37% used space

I have not rebooted esx1 to test if the problem still persists , because the system is in a production environment.

Both servers are not in a cluster ; they are independent of each other , and have different configurations.

Like I mentioned earlier , esx2, with lots of free space available , also has got this problem.

So what could be done to fix this problem on this server ? Any ideas?

Is there any way of forcing the system to commit the configurations into the disk?

BTW, only the most recent changes are discarded upon system reboot ; the old configurations are not affected.

This problem is pretty recent ; didn't have it before.

Thanks ,

Danny

[CameroonVM]

Hi Dan ,

I just rebooted the servers after "sync'ing" multiple times via command line and I don't have this prob anymore.

It looks like it was caused by full root ( "/" ) partition for esx1 and rapid shut down on esx2 .

Thanks,

Could u please , reply to that I can mark your previous response as "Correct Answer" ?

Thanks ,

Danny

[a2alpha · Accepted solution]

Really glad its sorted for you, as I said, i have seen the issue before at a customer site and you will never know that the root is full unless you look into the service console. It would be quite good to have this visible through the vi-client but in the mean time its just added to the monthly checks!

Thanks,

Dan

[CameroonVM]

It'd be great if this could be added to vSphere client.

Thanks again :smileygrin: :smileysilly: