I would like advise about how to replace a raid 0 lun. The lun is basically a
raid 0 lun which is from a single disk volume group / volume from our Flexline
380 unit. So every disk in the unit is a volume group/volume/lun mapped to the
host. We then let ZFS do the raid.
We have a lun now which has been getting read errors and basically the
underlying drive needs to be replaced. I''ve done this in the past but
as I remember it was cumbersome and didn''t go that smoothly so I would
like some advise about how to go about it. As I recall if you just fail the
disk via Santricity, the lun really doesn''t go offline from a host
point of via and hence ZFS still tries to write to the lun/disk. I believe I
either unmapped the lun to the host at which point ZFS kicked in a hot spare, or
I offlined the disk via ZFS first. Then at that point the drive can be replace
on the storage and the volume re-initialized then remapped back to the host. So
part of my question... does the above sound reasonable, or should I be doing
this differently? Also I''m a little unsure about how to get the
original lun back in operation and the spare back to a spare.
For example I have the following situation now:
raidz2 ONLINE 0 0 0
c10t600A0B800011399600007CF945E80E95d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FEEB45E8145Ed0 ONLINE 0 0 0
spare ONLINE 0 0 0
c10t600A0B800011399600007D2345E81075d0 ONLINE 0 0 0
c10t600A0B800011399600007CE145E80D4Dd0 ONLINE 0 0 0
c10t600A0B800011399600007D3F45E81157d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FF1145E817BEd0 ONLINE 0 0 0
c10t600A0B800011399600007D5D45E813EBd0 ONLINE 0 0 0
c10t600A0B8000115EA20000FE7D45E80CDEd0 ONLINE 0 0 0
c10t600A0B800011399600007C6145E808C7d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FE9945E80E6Ad0 ONLINE 0 0 0
c10t600A0B800011399600007C8B45E80A59d0 ONLINE 0 0 0
c10t600A0B800011399600007CA745E80B21d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FEB545E810D4d0 ONLINE 0 0 0
c10t600A0B800011399600007CD145E80CD7d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FED145E8129Cd0 ONLINE 0 0 0
c10t600A0B800011399600007CFB45E80EA5d0 ONLINE 0 0 0
c10t600A0B8000115EA20000FEED45E8146Ed0 ONLINE 0 0 0
spares
c10t600A0B800011399600007CE145E80D4Dd0 INUSE currently in use
c10t600A0B8000115EA20000FEE145E81328d0 AVAIL
c10t600A0B800011399600007D0B45E80F21d0 AVAIL
c10t600A0B8000115EA20000FEFD45E81506d0 AVAIL
c10t600A0B800011399600007D3545E81107d0 AVAIL
c10t600A0B800011399600007D5345E81289d0 AVAIL
c10t600A0B8000115EA20000FF2345E81864d0 AVAIL
c10t600A0B800011399600007D6F45E8149Bd0 AVAIL
I thought if I replace the original device with the same (but new) lun that
should work, but I get
the following:
zpool replace tank c10t600A0B800011399600007D2345E81075d0
c10t600A0B800011399600007D2345E81075d0
invalid vdev specification
use ''-f'' to override the following errors:
/dev/dsk/c10t600A0B800011399600007D2345E81075d0s0 is part of active ZFS pool
tank. Please see zpool(1M).
The above lun is the same as before, but the underlying disk on the storage was
replaced. Do I need to do something to this lun to make ZFS think it is a new
disk? Or should I be doing something different?
I now have another disk which has gone back so I need to fix my above situation
with the hot
spare first, and then go through the process again for my second failure.
The environment is S10U4, running on a x4600 with Flexline 380 storage units.
Tia,
David
This message posted from opensolaris.org
Addtional information:
It looks like perhaps the original drive is in use, and the hot spare is
assigned but not in use see below about zpool iostat:
raidz2 2.76T 4.49T 0 0 29.0K 18.4K
c10t600A0B800011399600007CF945E80E95d0 - - 0 0 2.46K
1.33K
c10t600A0B8000115EA20000FEEB45E8145Ed0 - - 0 0 2.46K
1.33K
spare - - 0 0 1.81K 1.33K
c10t600A0B800011399600007D2345E81075d0 - - 0 0 2.47K
1.33K
c10t600A0B800011399600007CE145E80D4Dd0 - - 0 0 0
1.33K
c10t600A0B800011399600007D3F45E81157d0 - - 0 0 2.47K
1.33K
c10t600A0B8000115EA20000FF1145E817BEd0 - - 0 0 2.46K
1.33K
c10t600A0B800011399600007D5D45E813EBd0 - - 0 0 2.46K
1.33K
c10t600A0B8000115EA20000FE7D45E80CDEd0 - - 0 0 2.47K
1.33K
c10t600A0B800011399600007C6145E808C7d0 - - 0 0 2.47K
1.33K
c10t600A0B8000115EA20000FE9945E80E6Ad0 - - 0 0 2.46K
1.33K
c10t600A0B800011399600007C8B45E80A59d0 - - 0 0 2.46K
1.33K
c10t600A0B800011399600007CA745E80B21d0 - - 0 0 2.47K
1.33K
c10t600A0B8000115EA20000FEB545E810D4d0 - - 0 0 2.47K
1.33K
c10t600A0B800011399600007CD145E80CD7d0 - - 0 0 2.47K
1.33K
c10t600A0B8000115EA20000FED145E8129Cd0 - - 0 0 2.47K
1.33K
c10t600A0B800011399600007CFB45E80EA5d0 - - 0 0 2.47K
1.33K
c10t600A0B8000115EA20000FEED45E8146Ed0 - - 0 0 2.46K
1.33K
spares
c10t600A0B800011399600007CE145E80D4Dd0 INUSE currently in use
So how do I get the hot spare out of the inuse state?
This message posted from opensolaris.org
Cindy.Swearingen at Sun.COM
2008-Mar-14 16:10 UTC
[zfs-discuss] Replacing a Lun (Raid 0) (Santricity)
David, Try detaching the spare, like this: # zpool detach pool-name c10t600A0B800011399600007CE145E80D4Dd0 Cindy David Smith wrote:> Addtional information: > > It looks like perhaps the original drive is in use, and the hot spare is assigned but not in use see below about zpool iostat: > > raidz2 2.76T 4.49T 0 0 29.0K 18.4K > c10t600A0B800011399600007CF945E80E95d0 - - 0 0 2.46K 1.33K > c10t600A0B8000115EA20000FEEB45E8145Ed0 - - 0 0 2.46K 1.33K > spare - - 0 0 1.81K 1.33K > c10t600A0B800011399600007D2345E81075d0 - - 0 0 2.47K 1.33K > c10t600A0B800011399600007CE145E80D4Dd0 - - 0 0 0 1.33K > c10t600A0B800011399600007D3F45E81157d0 - - 0 0 2.47K 1.33K > c10t600A0B8000115EA20000FF1145E817BEd0 - - 0 0 2.46K 1.33K > c10t600A0B800011399600007D5D45E813EBd0 - - 0 0 2.46K 1.33K > c10t600A0B8000115EA20000FE7D45E80CDEd0 - - 0 0 2.47K 1.33K > c10t600A0B800011399600007C6145E808C7d0 - - 0 0 2.47K 1.33K > c10t600A0B8000115EA20000FE9945E80E6Ad0 - - 0 0 2.46K 1.33K > c10t600A0B800011399600007C8B45E80A59d0 - - 0 0 2.46K 1.33K > c10t600A0B800011399600007CA745E80B21d0 - - 0 0 2.47K 1.33K > c10t600A0B8000115EA20000FEB545E810D4d0 - - 0 0 2.47K 1.33K > c10t600A0B800011399600007CD145E80CD7d0 - - 0 0 2.47K 1.33K > c10t600A0B8000115EA20000FED145E8129Cd0 - - 0 0 2.47K 1.33K > c10t600A0B800011399600007CFB45E80EA5d0 - - 0 0 2.47K 1.33K > c10t600A0B8000115EA20000FEED45E8146Ed0 - - 0 0 2.46K 1.33K > > spares > c10t600A0B800011399600007CE145E80D4Dd0 INUSE currently in use > > So how do I get the hot spare out of the inuse state? > > > This message posted from opensolaris.org > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Yes! That worked to get the spare back to an available state. Thanks! So that leaves me with the trying to put together a recommended procedure to replace a failed lun/disk from our Flexline 380. Does anyone have configuration in which they are using a RAID 0 lun, which they need to replace? Thanks, David This message posted from opensolaris.org