Hi,
This is a re-send. The first time I sent the message it was truncated
somewhere along the line.
- David Harris
Principal Engineer, DRH Internet Services
-----Original Message-----
From: David Harris [mailto:[EMAIL PROTECTED]]
Sent: Monday, December 21, 1998 1:01 AM
To: MOLNAR Ingo [[EMAIL PROTECTED]]; [EMAIL PROTECTED]
Subject: Raid error?
Hi,
Somehow my server ended up rebooting (how I don't know) and failed when
trying to fsck the filesystem on one of my raid devices. Have _no_ idea why
the system failed, but it did.
The error message on the fsck seems to indicate a problem reading the
underlying block device, /dev/md2. (It says "error writing block.") This is
why I'm reporting it here. I suspect a md kernel driver problem.
Here is the relevant stuff from my bootup console:
------------
Starting /dev/md2
(read) sda9's sb offset: 1534080 [events: 0000004d]
(read) sdb9's sb offset: 1534080 [events: 0000004d]
autorun ...
considering sdb9 ...
adding sdb9 ...
adding sda9 ...
created md2
bind<sda9,1>
bind<sdb9,2>
running: <sdb9><sda9>
now!
sdb9's event counter: 0000004d
sda9's event counter: 0000004d
md2: max total readahead window set to 128k
md2: 1 data-disks, max readahead per data-disk: 128k
raid1: device sdb9 operational as mirror 1
raid1: device sda9 operational as mirror 0
(checking disk 0)
(really checking disk 0)
(checking disk 1)
(really checking disk 1)
(checking disk 2)
(checking disk 3)
(checking disk 4)
(checking disk 5)
(checking disk 6)
(checking disk 7)
(checking disk 8)
(checking disk 9)
(checking disk 10)
(checking disk 11)
raid1: raid set md2 active with 2 out of 2 mirrors
md: updating md2 RAID superblock on device
sdb9 [events: 0000004e](write) sdb9's sb offset: 1534080
sda9 [events: 0000004e](write) sda9's sb offset: 1534080
.
... autorun DONE.
marking sb clean...
md: updating md2 RAID superblock on device
sdb9 [events: 0000004f](write) sdb9's sb offset: 1534080
sda9 [events: 0000004f](write) sda9's sb offset: 1534080
.
md2 switched to read-only mode.
------------
The md devices are up.
Checking filesystems.
Parallelizing fsck version 1.10 (24-Apr-97)
Checking all file systems.
----------------------------------
----------------------------------
[/sbin/fsck.ext2] fsck.ext2 -a /dev/md2
[/sbin/fsck.ext2] fsck.ext2 -a /dev/md1
[/sbin/fsck.ext2] fsck.ext2 -a /dev/md3
[/sbin/fsck.ext2] fsck.ext2 -a /dev/md4
/dev/md2 was not cleanly unmounted, check forced.
/dev/md4: clean, 11/841984 files, 26429/841376 blocks
/dev/md3: clean, 16/841984 files, 240609/841376 blocks
/dev/md1: clean, 339/52208 files, 28204/208640 blocks
Fix summary information? yes
/dev/md2: 58523/383520 files (3.2% non-contiguous), 942526/1534080 blocks
Error writing block 8 (Attempt to write block from filesystem resulted in
short write) while writing block bitmaps.
/dev/md2: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.
(i.e., without -a or -p options)
----------------------------------
I'll backup both of the underlying partitions for /dev/md2 and then play
around to see if I can get it working. Tell you if I get anything, but I
would appreciate any help or ideas.
- David Harris
Principal Engineer, DRH Internet Services
