Promise Fastrak ATA RAID 1 rebuild failed
Michael Shafae
cadaver at tucu.net
Tue Feb 10 22:39:58 PST 2004
Hi,
I have an intel motherboard, s845wd1, with an onboard Promise Fastrak
ATA RAID. Its running 4.8p10 on two mirrored Seagate 80 GB disks. It
lost the mirror reporting this:
ad6: hard error reading fsbn 140018943 of 0-127 (ad6 bn 140018943; cn
138907 tn
10 sn 57) trying PIO mode
ad6: DMA problem fallback to PIO mode
ad6: DMA problem fallback to PIO mode
ad6: DMA problem fallback to PIO mode
ad6: DMA problem fallback to PIO mode
ad6: hard error reading fsbn 140019007 of 0-127 (ad6 bn 140019007; cn
138907 tn
11 sn 58) status=59 error=00
ar0: WARNING - mirror lost
So following the handbooks steps at
<http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/raid.html> I
detached channel 3 (which has ad6), attached channel 3 and then
`atacontrol rebuild ar0`. An hour later it died with this message at
the prompt:
# atacontrol rebuild ar0
atacontrol: ioctl(ATARAIDREBUILD): Input/output error
On the console, I saw the following message:
ad4: hard error reading fsbn 75859712 of 0-255 (ad4 bn 75859712; cn
75257 tn 10 sn 26) trying PIO mode
ad4: DMA problem fallback to PIO mode
ad4: DMA problem fallback to PIO mode
ad4: DMA problem fallback to PIO mode
ad4: DMA problem fallback to PIO mode
ad4: DMA problem fallback to PIO mode
ad4: read error detected (too) late
Running `atacontrol status ar0` returned:
ar0: ATA RAID1 subdisks: ad4 ad6 status: REBUILDING 48% completed
I waited 10 minutes or so and ran the command again and got the same
output.
I rebooted the system and it came up ok but with these messages on the
console:
ata0-slave: ATAPI identify retries exceeded
ar0: 76293MB <ATA RAID1 array> [9726/255/63] status: DEGRADED subdisks:
0 READY ad4: 76319MB <ST380021A> [155061/16/63] at ata2-master UDMA100
1 SPARE ad6: 76319MB <ST380021A> [155061/16/63] at
ata3-master UDMA100
acd0: CDROM <CD-540E> at ata0-master PIO4
Mounting root from ufs:/dev/ar0s1a
swapon: adding /dev/ar0s1b as swap device
Automatic boot in progress...
/dev/ar0s1a: FILESYSTEM CLEAN; SKIPPING CHECKS
/dev/ar0s1a: clean, 78595 free (123 frags, 9809 blocks, 0.1%
fragmentation)
/dev/ar0s1f: FILESYSTEM CLEAN; SKIPPING CHECKS
/dev/ar0s1f: clean, 515272 free (720 frags, 64319 blocks, 0.1%
fragmentation)
/dev/ar0s1g: FILESYSTEM CLEAN; SKIPPING CHECKS
/dev/ar0s1g: clean, 31729380 free (80444 frags, 3956117 blocks, 0.2%
fragmentation)
/dev/ar0s1e: FILESYSTEM CLEAN; SKIPPING CHECKS
/dev/ar0s1e: clean, 2975836 free (1076 frags, 371845 blocks, 0.0%
fragmentation)
If I understand this correctly, ar0 is now basically ad4. ad4 was the
one that had the read error so ad6 couldn't be rebuilt. ad6 had a
similar error and that's why the mirror was lost.
I would like to rebuild the mirror, but I don't see how I can do this
without re-installing the entire system. Has anyone had any similar
experience or some insight on how I can get out of this bind?
Thanks,
Michael
More information about the freebsd-questions
mailing list