Hello, over the weekend I planned to remove a LUN that was wasting space on our SAN.
I vmotioned off the few VMs that were on it, and then removed the datastore.
I presumed that after this I could take the lun offline (using the SAN console) and then refresh the storage adapters, and the lun would be gone.
I did this, and half the the VMs running on our servers kind of went into this disconnected state. They werent powered off or anything, but I couldnt ping them or manage them at all..... I put the lun back to the online state, and everything started working again.
I then tried to delete the "static discovery" record of the lun in the iSCSI settings of each VM... The same thing happened, and even one of our ESXi boxes then kind of went offline, HA fell over and all hell broke loose.
I have since reverted all of this and everything is now working, but I still need to remove this lun.
Anyone have any idea why removing an empty lun is causing sooo much drama?
Did you vmotion or storage vmotion? If it was vmotion the virtual machines owuld not have moved form the LUN - use storage vmotion to move the vms off to another LUN and then from the storage option on the configuration tab for any of the ESX hosts that access this LUN select the VMFS datastore and remove datastore - you then should be able to remoce the LUN
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful
Yes, it was storage vmotion I used. I browsed the datastore to ensure it was empty.
I did also remove the datastore. The only thing left in vsphere that refers to the lun is its iSCSI target in the properties of the iSCSI initiator (under static discovery tab)..
But as I said, when I remove this, half the servers become disconnected.... Uunable to ping or open console... and even one of the ESX boxes became disconnected from vshpere, I had to go manage it directly... It seems as though there is some kind of dependancy on the lun in vsphere somewhere.
once you have removed the VMFS datastore - you can stop presenting the LUN from the SAN - once that is done you will need to go back to the vSphere client and rescan the SAN and the LUN should no longer be identified on the ESX host -
If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful