From: Iain Barker (ibarker@aastra.com)
Date: Wed May 14 2003 - 09:28:49 EDT
I got no suggestions from this list, which usually implies I am trying
either to do something wrong, or what I am trying isn't possible.
HP support first suggestion was to restore from backup - would I be trying
to do this, if we had a backup??? But HP were helpful in sugggesting that if
LSM was removed from the disk then it might be possible to run advfs salvage
more effectively on the raw partition.
>From HP:
Not being able to get into multi-user mode from either
disk limits what we can perform. I don't think using
volrootmirror would work because LSM would detect that it
has partially mirrored volumes and halt.
I am seeing greater value in salvage as everything in our data
base indicated that with your error a restore is the only
solution. Removing LSM uncomplicates things and allows more
direct access to the disk.
That is exactly what I did, and the system is now recovered. For referemce
of anyone else who has similar problems in future, the sequence is below
(rz9=good boot disk, rz11=missing mirror).
remove the stale mirror plexes from LSM : volplex dis rootvol-02, voledit
-r rm rootvol-02
remove the LSM encapsulation of the root disk : volunroot -a
recreate the advfs domains with advscan : /sbin/advfs/advscan -r rz9
salvage the usr_domain data to another disk : cd /tmp ; /sbin/advfs/salvage
rz9
delete the usr_domain : rmfdmn usr_domain
recreate the usr_domain and usr fileset : mkfdmn usr_domain /dev/rz9e,
mkfset usr_domain usr
mount the new fileset as /usr : mount -t advfs usr_domain#usr /usr
copy back the saved data from spare disk to /usr : cp -Rp /tmp/usr/* /usr
re-encapsulate the root disk : volencap -a rz9
insert a new disk in rz11 and add a blank disklabel : disklabel -z rz11;
disklabel -wr rz11
re-mirror rz9 to rz11 and update rz11 partition table : volrootmir -a rz11
- Iain
-----Original Message-----
From: Iain Barker [mailto:ibarker@aastra.com]
Sent: Tuesday, 13 May, 2003 08:31
To: tru64-unix-managers@ornl.gov
Subject: Advfs mount error on broken LSM mirror : E_CANT_FIND_LOG_END
This one seems to be a common problem on the list, and even though I find a
lot of people having the problem, nobody ever summarised a solution.
System as LSM encapsulated root disk, which includes two advfs domains: root
and usr
LSM drive is mirrored between two physical disks and the system shutdown
cleanly (i.e. both mirror drives are complete, neither is stale at
shutdown).
When the system is brought back up (a few months later) one of the mirrored
drives has failed, and is not probed by SRM/kernel, hence only half of the
LSM mirrorred volume is present.
That should be enough to mount? The mirror isn't there but the LSM volume
still exists and all the data is there on the one good copy of the mirror.
But ADVFS fails to mount on the LSM volume and complains that it cannot find
the logfile for the /usr filesystem. Interestingly the / root filesystem is
set up identically and works without any problem with only one half of the
mirror present.
Any ideas for a solution to this? The only rather drastic solution I can
think of is to use salvage to pull any partial files off /usr to a new disk.
But that isn't likely to result in a 100% stable system.
Because nothing actually happened to the 'good' disk, I feel there must be
some way to fix the problem within LSM/Advfs, but don't have enough
knowledge of those products.
A little help?
thanks.
Iain
starting LSM
Checking local filesystems
/sbin/ufs_fsck -p
Mounting / (root)
user_cfg_pt: reconfigured
Mounting local filesystems
exec: /sbin/mount_advfs -F 81920 root_domain#root /
root_domain#root on / type advfs (rw)
/proc on /proc type procfs (rw)
exec: /sbin/mount_advfs -F 16384 usr_domain#usr /usr
advfs_logger: pg #510 is bad; truncating log
advfs_logger: pg #509 is bad; truncating log
advfs_logger: pg #508 is bad; truncating log
advfs_logger: pg #507 is bad; truncating log
advfs_logger: pg #506 is bad; truncating log
advfs_logger: pg #505 is bad; truncating log
advfs_logger: pg #504 is bad; truncating log
advfs_logger: pg #503 is bad; truncating log
advfs_logger: pg #502 is bad; truncating log
advfs_logger: can't find log end
bs_bfdmn_activate: can't open domain ftx log, tag = 0xfffffff7.0x0:
E_CANT_FIND_LOG_END (-1077)
usr_domain#usr on /usr: I/O error
Unhelpfully, 'man advfs_err' says for this log:
E_CANT_FIND_LOG_END There is an error in finding the end of the log
Previous list postings of this same error (can find no summaries) here:
http://www.ornl.gov/cts/archives/mailing-lists/tru64-unix-managers/1996/04/m
sg00201.html
http://www.ornl.gov/cts/archives/mailing-lists/tru64-unix-managers/1998/03/m
sg00285.html
http://aa11.cjb.net/tru64_unix_managers/2001/05/0413.html
http://aa11.cjb.net/tru64_unix_managers/2000/03/0370.html
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:18 EDT