I have two ESX 3.5 servers in an HA Cluster. I had a problem conversion that I stopeed and was unable to delete the data from the datastore. I thus removed the LUN at the SAN level. When I rescanned my IScsi adapters ESX1 showed a DEAD path LUN. ESX2 after rescan did not and showed-up properly. I have rescaned ESX1 twice to no avail. I subsequently added a new LUN and both ESX1 and 2 picked-up the new LUN but ESX1 is +1 on number of LUN indicated due to the Dead path LUN.
ISCSI NAME AND PATH: DEAD
Also when I went to add to storage the 'new' LUN it also showed the Dead Path Lun.
How do I remove the Dead Path Lun from ESX1?
Well, I know it's too late, but you should never do that!
If you want to remove a LUN from a host, first unpresent it, then rescan. If everything is ok you can delete the LUN, if not you can re-present it.
However, it seems there is either a file reference (VM, ISO, ...) or a running process which points to the deleted datastore. You will have to check your inventory for anything pointing to the deleted datastore. If you can't find anything you could either search for processes or reboot the host.
I've encountered the same problem on ESX4.1 and an esxcfg-rescan -d vmhba0 also resolved this for me.
How can I do this via esxcfg-rescan or vicfg-rescan from a vMA? from the vMA the options for vicfg-rescan are different. I don't want to have to logon to the hosts individually.
The rescans via the VI client do not work.
It seems as though these rescans (VI client or vicfg-rescan) look for new devices and mount it if it contains a VMFS volume. The esxcfg-rescan -d from the console is able to delete the path from memory if the LUN has been removed.
Can anyone verify if this is true?