Removed LUN still showing on rescan..

[Albert Smith_1]

Hello Fellow Admins,
I am stumped on a problem.

I have a RHEL5.2 64 bit hosts that had a total of 3 luns

luns 0 1 2

Lun 0 was unassigned from the host and deleted from the array. However on a FC rescan LUN 0 is still being detected however I/O errors are generated as the disk is not assigned to the machine.

I have removed the devices file and done the single scsi remove from the /proc/scsi/scsi but on every reboot and rescan the lun is stil detected.

We don't have anything trying to access it.

Any suggestions on what I am doing wrong here?

Thanks,

ACS

[Jupinder Bedi · Accepted solution]

If it was a multipath lun you can reomove it by using following command

multipath -f

Please see this RedHat Knowledge Base document:
http://kbase.redhat.com/faq/docs/DOC-3942

Summary: to remove a device:
echo 1 > /sys/block//device/delete

and than rescan it

Good luck

[Matti_Kurkela]

FibreChannel is compatible with SCSI, and all SCSI devices must have LUN 0, since it's what is used for SCSI device identification ("inquiry" in SCSI terminology).

Because of this, you cannot really un-present LUN 0 while still presenting other LUNs: completely un-presenting LUN 0 would cause the device/LUN detection to fail, and you would not be able to use the other LUNs. Most SAN management programs "know" this and act accordingly.

MK

[Albert Smith_1]

I had a funny feeling that is what it was. As I have not seen this on other OS's it had my scratching my head a bit.

I will take an outage at our next window and fix it.

Thanks for the info!

-ACS

[Albert Smith_1]

I need to redeploy with a LUN 0.