Removing dead paths in ESXi 4.1

[devs159]

Hi,

We currently have fibre storage and when a LUN was removed on the SAN it's still being picked up by the hosts as a dead path. The volume has been unmapped/deleted from the SAN. Unfortunately, you can't manage the hosts as they've disconnected but the VM's are still up.

Re-scanning on the service console lists the dead path.

(Device is in use by worlds:)

Any suggestions for removing this without rebooting the ESXi host?

[MauroBonder]

Hi Devs,

Welcome.

you just tried remove lun manually ? or the lun not shown in Host ?

you can try remove this path using console esxcfg-rescan -d vmhba

Please, don't forget the awarding points for "helpful" and/or "correct" answers.

Mauro Bonder - Moderator

[MauroBonder]

moved from VMware ESXi™ 4 to VMware vSphere™ Storage

[devs159]

Hi Mauro,

The LUN was removed manually by our storage team. The hosts are disconnected and the datastore is listed as inactive.

When I run esxcfg-rescan -d vmhba it lists the dead path and this is what I'm triyng to remove/kill.

Device is in use by worlds:

World     # of Handles Name

4160                       1 idle0

I've traced it back to a number of VM's which were on the LUN. I may need to kill each of the VM's that were on the LUN removed and see if it gets our hosts back then.

[MauroBonder]

to solve problem "show how inactive" - http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101508...

Now, if your SAN zoning staff removed this mapping should disappear the way dead. A quick solution to remove this way is to reboot. But make sure that this configuration of SAN (Zone) was removed.

Please, don't forget the awarding points for "helpful" and/or "correct" answers. 

Mauro Bonder - Moderator

[devs159]

Afraid, we can't restart just yet as there's over 50 VM's on each host affected. I managed to track down the World ID for 2 out of the 4 VM's on the datastore (inactive) and once removed through s/console the hosts reconnected fine. I'm still having a problem with two hosts and can't find the World ID for the two VMs.

There's nothing listed when I run vm-support -x. Also, can't see the VM via ps | grep vmx. I was using kb 1014165 but still need to find the two VMs and kill the associated processes.

[MauroBonder]

try run services.sh restart

and read this kb, maybe help http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100434...

Good luck

Please, don't forget the awarding points for "helpful" and/or "correct" answers. 

Mauro Bonder - Moderator

[mrcpld]

Try to:

esxcfg-scsidevs –c   (find the dead naa device)

esxcfg-scsidevs –o naa.....      (put offline the device)

than rescan

that's all

Bye

MP

[saibala]

this is a great suggestion to use esxcfg-scsidevs -o followed by esxcfg-rescan

this worked as a panacea to our APDs/DRS / many other issues

please award points (atleast 10) to mrcpld