Storage problems on ESX 4.1 when a LUN disappears

[Aristizabal]

Hello,

I have been having problems on my ESX 4.1 servers whenever I have to take offline a Datastore to run any maintenance at the storage level.

I remember to have experienced similar issues when I upgraded from ESX 3.5 to 4.0, and VMware released a patch to address this.

VMware ESX 4.0, Patch ESX400-200912401-BG: Updates vmkernel, vmklinux, tools, CIM, and perftools (1016291)

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101629...

But for ESX 4.1 there are no patches available, so I am wondering if I am the only one experiencing this.

On my case, the moment I remove the LUN the ESX server starts having problems, like losing access through the vSphere Client and failing to execute any vSphere CLI commands.

Any clues on how to prevent this behavior on the ESX server is highly appreciated.

Thanks,

Juan.

[Aristizabal]

Sorry for replying to myself, but here is an update from VMware support about the issue for everyone's benefit.

"There have been some issues in 4.0 and 4.1 due to the new storage architecture. These have been addressed with patches which have been rolled in with 4.1 but won't be completely mitigated until the next major release of ESX. ESX 3.x just happened to handle the case where storage was removed from underneath it better but it was still not recommended to just remove a LUN that contains a valid datastore."

I was referred to the following KB, which explains how to unpresent a LUN containing a valid datastore to the ESX server:

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101508...

The KB applies to 4.0 but I have tested the procedure on 4.1 and works fine. I got rid of my storage problems by masking the LUN before it is removed from the ESX.