badblocks(8) while fs mounted

[Sven Heinicke <sven@nec-labs.com>]

I want to test a partition, 18 lvm'd 80G disks to an ext3 1.2T
partition, and make it return:
 
Apr 28 06:25:07 ds1 kernel: hds: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Apr 28 06:25:07 ds1 kernel: hds: dma_intr: error=0x40 { UncorrectableError }, LBAsect=31821782, sector=31821776
Apr 28 06:25:07 ds1 kernel: end_request: I/O error, dev 5b:00 (hds), sector 31821776

errors when a disk has something wrong with it of the types these errors
indicate.  So, I want to run mke2fs -c, but I can't take the disks off
line.  I thought about doing a dump(8) of the partition to /dev/null but
I want to check all the blocks, not just where there is data.  Is
dd(1)ing it to /dev/null ok, is there a better way?


That was my question, now the background:

I can't take these disks off line, but they only ever get used once a
quarter or so.  So they sit idle for months, somebody logs and the disk
starts returning errors.  Now, this disk probably went bad weeks ago,
but it wasn't accessed until now, so no errors until now.  I want to
catch the errors before a user does, via a weekly cron job.  When these
disks get used, they don't seem to have problems, so hopefully cron job
will stop these errors too.

  Sven
_______________________________________________
LinuxManagers mailing list - http://www.linuxmanagers.org
submissions: LinuxManagers@linuxmanagers.org
subscribe/unsubscribe: http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers