Bad blocks in raid5 disk
- From: Schraalhans Keukenmeester <schraalhans.keukenmeester@xxxxxxxxxxxx>
- Date: Wed, 29 Mar 2006 21:45:02 +0200
I have the following situation:
3 harddisks, 2 P-ATA, 1 SATA. All 160 GB disks, not relevant I think)
/dev/hda, dev/hdb, /dev/sda
On each disk a 20GB partition (type fd) that is used in a software raid5 array /dev/md0. (/dev/hda5, /dev/hdb3, /dev/sda2)
My / is mounted on this raid volume.
Suddenly I got swamped in messages indicating disk problems on /dev/sda2
(Driveseek, Read errors)
(Temporarily stopped mdmonitor and auditd, as my system became almost totally unresponsive, as each error produces another message.)
Failed the /dev/sda2 device (mdadm /dev/md0 --fail /dev/sda2 --remove /dev/sda2)
Tried to fsck the partition but can't. There seems to be no fstype suited for checking a 'raid auto' partition.
Ran badblocks on the device and found bad blocks.
I'll buy a new disk asap.
For the time being : (how) Can I use fsck on this fd- partition to map out the bad sectors ? Since the SATA-device is treated 'as scsi', does the driver itself take care of the bad block relocation? How can I best proceed, given the fact the drive cannot be replaced until next week or so ? I found some wish-list for mdadm, so I assume this driver isn't fully developed yet. I'd rather have tools to 'repair' the thing from within the raid array. Is that possible ??
Thanks
Sh.
.
- Follow-Ups:
- Re: Bad blocks in raid5 disk
- From: Grant
- Re: Bad blocks in raid5 disk
- From: John-Paul Stewart
- Re: Bad blocks in raid5 disk
- Prev by Date: Re: ioctl(...) with raw socket problem?
- Next by Date: Re: Bad blocks in raid5 disk
- Previous by thread: How to make a BASH script echo the commands?
- Next by thread: Re: Bad blocks in raid5 disk
- Index(es):
Relevant Pages
|