VMware Cloud Community
albrandwood
Contributor
Contributor
‎04-19-2014 08:50 AM

VMware ESXi5.5-1474528 and iSCSI not starting on boot.

I have two ESXi 5.5 1474528 build VM machines (Dell CS-24TY).  Both are connected using software iscsi controller to a Promise iSCSI RAID.

The setup works almost flawlessly ... Almost, because neither will automount the iSCSI on boot.

In both cases, I have to manually mount using esxcfg-volume -m <UUID> ... after that, they work without any issues.

Any help? Suggestions?

Tags (4)
0 Kudos
3 Replies
a_p_
Leadership
Leadership
‎04-19-2014 09:52 AM

Not sure, but maybe worth checking whether the LUN is detected as a snapshot LUN (see http://kb.vmware.com/kb/1011387)

You may also want to use the esxcli (mentioned in the KB article) command on ESXi 5.5 to persistently mount the LUN.

Btw. for the esxcfg-volume command, it's important whether "-m" or "-M" is used for mounting LUNs.

André

0 Kudos
albrandwood
Contributor
Contributor
‎04-19-2014 12:09 PM

OK ... the vmkernel.log on both machines :

2014-04-19T14:59:40.584Z cpu10:33258)LVM: 8349: Device eui.22cd000155a3fe57:1 detected to be a snapshot:

2014-04-19T14:59:40.584Z cpu10:33258)LVM: 8356:   queried disk ID: <type 1, len 44, lun 0, devType 0, scsi 0, h(id) 10402944932408356314>

2014-04-19T14:59:40.584Z cpu10:33258)LVM: 8363:   on-disk disk ID: <type 1, len 44, lun 0, devType 0, scsi 0, h(id) 15507285305951941117>

So I guess they are seeing it as a snapshot ... the LUN ID appears correct (eui.22cd000155a3fe57:1) on both machines,

The Device ID is the same on both machines (eui.22cd000155a3fe57) and formated as VMFS 5.60

I will try : esxcli storage vmfs snapshot mount -u <VMFS UUID> Ithe next time I reboot one of the servers...

Is there a "correct" procedure to remove the "snapshot" status from the drive?

From what i can see, the procedure is:

  1. Shutdown all Guests
  2. deregister the Guests
  3. Place both Servers in Maintenance mode
  4. unmount the iSCSI from both servers
  5. resignature the iSCSI using on server
  6. remount the iSCSI on both servers
  7. Remove both servers from maintenance mode.
  8. reregister the Guests
  9. power on all Guests

Am I right in assuming that performing a resignature is not destructive to data already on the drive, it's just changing the UUID of the drive ?

And therefor, in theory, the "queried disk ID" and the "on-disk disk ID"  in the vmkernel.log output above would then be the same IDs ?

Thanks in advance.

@shley

0 Kudos
a_p_
Leadership
Leadership
‎04-19-2014 12:31 PM

Do the LUNs currently show up as snapshot LUNs if you run:

esxcli storage vmfs snapshot list

Without knowing what was causing the LUNs as being detected as snapshot LUNs, it's hard to say what needs to be done. I'd actually double check how the LUNs are presented to the hosts (on the storage system) and - if there's nothing unusual - persistently mount the LUN next time. Since the non-persistent mount (lower-case "-m") worked for you, I would not expect any issues with the persistent mount.


André

0 Kudos