http://oss.sgi.com/bugzilla/show_bug.cgi?id=186
c.pascoe@xxxxxxxxxxxxxx changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |c.pascoe@xxxxxxxxxxxxxx
------- Additional Comments From c.pascoe@xxxxxxxxxxxxxx 2002-10-20 19:07
-------
I also experienced the shutdown filesystem described in this job and also
previously by Ian Hardy (difference being that Ian's was on a MD device),
after experiencing an unclean shutdown.
I have the same configuration (software wise) as described in this message.
My machine is also serving up files via NFS as per this job. It also has ECC
memory and has reported no ECC errors.
History before my crashes:
* System had been running kernel from 24/5/2002 plus critical fixes merged
by me (booted on 12/6/2002)
* On the evening of 2/10/2002 AEST I ran a firmware update on the machine
the machine crashed (whilst it was live, in preparation for restarting it -
I've done this several times before successfully though!)
* The machine restarted into the kernel that I had built and booted on an
identical box, compiled from CVS taken on 2/10/2002 AEST. This kernel was
CVS up to either modid 2.4.x-xfs:slinx:128691a or possibly modid
2.4.x-xfs:slinx:128671a. The machine performed recovery on the filesystem
at this time successfully.
* The next day I realised I hadn't merged the LVM 1.0.5 and VFS-lock code
in, so pulled in the changes from CVS that were made overnight and added the
LVM patches. At this point in time, the kernel was CVS to modid
2.4.x-xfs:slinx:128918a plus LVM 1.0.5 and LVM vfs-lock patches. I built
this kernel and put it in place to boot; did not test it on another machine.
* I then managed to lock up a RAID controller whilst trying to create a new
container and the machine restarted itself as a result of this. (No, I am
not having a good week!!!). I believe all commands to the array from this
time were abandoned.
* The machine restarted, and did NOT report that it performed recovery on
the disk. It completed the boot process, then eight minutes later I have
recorded on the serial console:
xfs_force_shutdown(lvm(58,0),0x8) called from line 1041 of file xfs_trans.c.
Return address = 0xf89dcac5
Corruption of in-memory data detected. Shutting down filesystem: lvm(58,0)
At this point, the watchdog on the machine restarted it again, it booted the
kernel that had been running forever (24/5/2002+), performed recovery and
everything has been fine since.
Since this failure, I have taken the filesystem offline and run xfs_check
across it - it found no problems with the on-disk data.
Regards,
Chris
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
|