This seems to be a limitation in the Linux kernel. Use the command
> cat /proc/scsi/scsi
To see what SCSI devices the system thinks is present. For the Xraid these look like:
Host: scsi3 Channel: 00 Id: 00 Lun: 01 Vendor: APPLE Model: Xserve RAID Rev: 1.50 Type: Direct-Access ANSI SCSI revision: 05
A quick way to list all disks is:
> cat /proc/scsi/scsi | grep APPLE
There should be one entry for each Xraid device. If the Xraid disks have been “sliced” then each slice appears as a unique device. If the Xserve disks are not present (or not enough are present) log on as root and run the script:
> rescan-scsi-bus.sh
This should “find” the missing disks.
Some one (or some process) has “cd”ed into the filesystem you want to unmount. First check your login windows for obvious offenders. Remember if you had “su”ed to root, then the original process may be the problem. Often the problem is people who have logged in remotely (e.g. for fringe testing). The best way to find this is
> lsof | grep xraid
This lists the processes which have a file open in the data diretories. Note the PID of processes identified above and use kill
(or kill -9
if necessary)
to remove the process. Don't “kill” processes blindly - check each one is not a vital process (such as vsib_record or recorder_server).
If the errant procoes is recorder_server
then run (as vlbi)
> /etc/init.d/recorder_server restart
You may have to login as root for lsof
to work. As a normal user, lsof
does not list root processes.
Log on as root and retry