Ghost volumes

limbo's picture

Carbon copy cloner, which had been doing a clone of the xsan volume nightly started to complain that there were multiple dries with the same unique identifier as the source. After looking I find that diskutil list shows multiple instances of the volume. I've restarted the backup system and unmounted and mounted the xsan volume but only went from two phantoms to 3. I haven't been able to shutdown the entire fabric yet because I haven't been able to schedule downtime for the server.

Anyone seen this before? Any tricks to resolving this?

Here is a portion of the diskutil output:

/dev/disk3

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan_Component *2.0 TB disk3
/dev/disk4

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan_Component *18.0 TB disk4
/dev/disk5

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan_Component *18.0 TB disk5
/dev/disk6

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan MyVolume *36.0 TB disk6
/dev/disk7

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan MyVolume *36.0 TB disk7
/dev/disk8

  1. : TYPE NAME SIZE IDENTIFIER

0: Apple_Xsan MyVolume *36.0 TB disk8

This is on 10.7.2

wrstuden's picture

Use XsanPrefs or XsanAdmin to unmount a volume. The extra volumes are puppet devices not cleaned up from the previous mounts. When you use XsanAdmin or XsanPrefs, xsand will unmount the volume and clean up the puppet device. If you use the finder, the cleanup step gets missed.

Reboot the client and they will all go away. Your Fabric and LUNs are fine.

no_xvi's picture

be worth going to /Volumes/ and deleting the ghost too.

Will also ensure that when you restart the volume mounts.

Regards,
SJ

wrstuden's picture

no_xvi wrote:
be worth going to /Volumes/ and deleting the ghost too/quote
This situation does not involve ghost volumes in /Volumes. The ghosts are only in DiskUtility (and disk arbitration). So don't go deleting things.