zfs discuss - Dec 2009 - part of active zfs pool error message reports incorrect decive

I am seeing this issue posted a lot in the forums:

A zpool add/replace command is run, for example:

zpool add archive spare c2t0d2

invalid vdev specification
use ''-f'' to override the following errors:
/dev/dsk/c2t1d7s0 is part of active ZFS pool archive. Please see zpool(1M).

(-f just says: the following errors must be manually repaired:)

Also, when running format and selecting this disk, the same error message is
printed about the c2t1d7s0 device.

In my case, I am running the current arcmsr driver for an Areca 1680 controller.
The disks are running in pass-through mode.  I have had issues with too many I/O
errors causing zfs fault devices.  Many months of working on these issues has
caused confusion in zfs because (I believe) at times physical devices have
mapped to different solaris disk devices.

Finally I believe I might have resolved the I/O problems and have a healthy
zpool:
        NAME        STATE     READ WRITE CKSUM
        archive     ONLINE       0     0     0
          raidz1    ONLINE       0     0     0
            c2t1d4  ONLINE       0     0     0
            c2t1d5  ONLINE       0     0     0
            c2t1d0  ONLINE       0     0     0
            c2t1d1  ONLINE       0     0     0
            c2t0d1  ONLINE       0     0     0
          raidz1    ONLINE       0     0     0
            c2t0d3  ONLINE       0     0     0
            c2t0d7  ONLINE       0     0     0
            c2t1d3  ONLINE       0     0     0
            c2t0d5  ONLINE       0     0     0
            c2t0d4  ONLINE       0     0     0
          raidz1    ONLINE       0     0     0
            c2t0d0  ONLINE       0     0     0
            c2t0d2  ONLINE       0     0     0
            c2t0d6  ONLINE       0     0     0
            c2t1d6  ONLINE       0     0     0
            c2t1d7  ONLINE       0     0     0

Yet I was still unable to add in my last drive (16 SAS drives, 15 are listed
above) as a spare.

I have tried:
export/importing the zpool.
rewriting the label to c2t1d2 as EFI
dd''ing the first and last 1M of c2t1d2

I keep getting that c2t1d7s0, a completely different device, is part of an
active zfs pool.  I don''t know for sure but I guess at some point this
physical drive was c2t1d7 in solaris, but either way, who cares now?

For the record, I have come up with a work-around I came up with, but this
doesn''t answer the question.

I deleted the pass-through drive in question in the RAID controller interface
then recreated it using a different SCSI ID and LUN.
-- 
This message posted from opensolaris.org

I''d agree with export/import *IF* the drive should be good, however I
have a drive that was pulled from the pool a long time ago (to flash drive the
drive) - The data on it is useless.  exporting/importing would cause either a)
errors, b) scrub to need to be run which should overwrite it completely.

This seems like some simple method to say FORCE this drive out of the pool.

Steve Radich
www.BitShop.com
-- 
This message posted from opensolaris.org