I've had my RAID0 array (hdb and hdd) working well for a couple of years...
until tonight. (Of course just a few hours after my large backup hard drive
(located in another machine) failed!!!)
the kernel (fedora core 1 2.4.22) was accessing hdc (the ide cd-writer) when
a problem occurred with that (irq timeout, ATAPI resets, etc, etc) when I
put a cdr into it. Soon afterwards, some hdd status timeouts occurred
(I'm guessing it was too busy trying to contact hdc)...
and ide1 reset timed-out errors
rather stuck, I shutdown -r now.
a few I/O errors occured on hdd but eventually it rebooted
On reboot, however, my raid0 was unfunctional!
Dec 30 21:34:32 zazu kernel: md: md driver 0.90.0 MAX_MD_DEVS=256, MD_SB_DISKS=27
Dec 30 21:34:32 zazu kernel: md: Autodetecting RAID arrays.
Dec 30 21:34:32 zazu kernel: [events: 00000130]
Dec 30 21:34:32 zazu kernel: [events: 000000cc]
Dec 30 21:34:32 zazu kernel: md: autorun ...
Dec 30 21:34:32 zazu kernel: md: considering hdd1 ...
Dec 30 21:34:32 zazu kernel: md: adding hdd1 ...
Dec 30 21:34:32 zazu kernel: md: adding hdb1 ...
Dec 30 21:34:32 zazu kernel: md: created md0
Dec 30 21:34:32 zazu kernel: md: bind<hdb1,1>
Dec 30 21:34:32 zazu kernel: md: bind<hdd1,2>
Dec 30 21:34:32 zazu kernel: md: running: <hdd1><hdb1>
Dec 30 21:34:32 zazu kernel: md: hdd1's event counter: 000000cc
Dec 30 21:34:32 zazu kernel: md: hdb1's event counter: 00000130
Dec 30 21:34:32 zazu kernel: md: superblock update time inconsistency -- using the most recent one
Dec 30 21:34:32 zazu kernel: md: freshest: hdb1
Dec 30 21:34:32 zazu kernel: md: kicking non-fresh hdd1 from array!
Dec 30 21:34:32 zazu kernel: md: unbind<hdd1,1>
Dec 30 21:34:32 zazu kernel: md: export_rdev(hdd1)
Dec 30 21:34:32 zazu kernel: md0: former device hdd1 is unavailable, removing from array!
Dec 30 21:34:32 zazu kernel: kmod: failed to exec /sbin/modprobe -s -k md-personality-2, errno = 2
Dec 30 21:34:32 zazu kernel: md: personality 2 is not loaded!
Dec 30 21:34:32 zazu kernel: md :do_md_run() returned -22
Dec 30 21:34:32 zazu kernel: md: md0 stopped.
Dec 30 21:34:32 zazu kernel: md: unbind<hdb1,0>
Dec 30 21:34:32 zazu kernel: md: export_rdev(hdb1)
Dec 30 21:34:32 zazu kernel: md: ... autorun DONE.
So it seems I still have 2 undead drives, but hdd is a little show with the
event counter?!? (i.e. no eents happened while the kernel coun't contact it?)
perhaps a few files should be impossible to recover but I would suspect that the
system itself should be recoverable if appropriate information can be copied
accross from hdb? Or do I need to recreate the entire array from scratch and
slightly-too-old backups-or-my-backups?
Is it possible to just copy appropriate superblock entries accross from hdb
somehow on this RAID0 system. Similar recovery methods seem possible with
RAID5 from what I read, but I cannot figure out exactly the commands to use.
(do I just call hdd as being failed in /etc/raidtab, or something like mkraid --force?)
thanks in advance for any suggestions,
--
Andrew Mitchell andrewm@cse.unsw.edu.au
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
