hard drive causing trouble with software raid array

machs_fuel · 07-15-2006, 02:51 PM

i'm setting up a file server with an older asus mother board and a promise ide card.
i've got debian installed on the primary master - hda, and there are 3 320 gig ide drives on the ide controller - hde hdf and hdh.

I've got mdadm installed and created the array with the following command

mdadm --create --verbose /dev/md0 --level=5 --raid-devices=3 /dev/hde /dev/hdf /dev/hdh

and then formatted with

mkfs.xfs /dev/md0 -L media_drive

i'm having a lot of

end_request: I/O error, dev 21:40 (hdf), sector 312572151
or
end_request: I/O error, dev 22:40 (hdh), sector 312572047

errors coming up which causes the array to drop hdf and run in degraded mode.

I'm not really sure how to proceed at this point. all three of the drives are brand new so i dont think that they are the problem. the IDE card was used so i suppose it could be the source. The mobo is probably the weakest link... it seems that the secondary ide channel is dead, but other wise works fine

machs_fuel · 07-15-2006, 02:54 PM

this is a dump of dmesg

Code:

 (hdh), sector 312572019
end_request: I/O error, dev 22:40 (hdh), sector 312572056
end_request: I/O error, dev 22:40 (hdh), sector 312572057
end_request: I/O error, dev 22:40 (hdh), sector 312572058
end_request: I/O error, dev 22:40 (hdh), sector 312572059
**snip**
end_request: I/O error, dev 22:40 (hdh), sector 312572061
end_request: I/O error, dev 22:40 (hdh), sector 312572062
end_request: I/O error, dev 22:40 (hdh), sector 312572063
md: updating md0 RAID superblock on device
md: (skipping faulty ide/host2/bus1/target1/lun0/disc )
md: ide/host2/bus0/target1/lun0/disc [events: 00000003]<6>(write) ide/host2/bus0/target1/lun0/disc's sb offset: 312571136
md: recovery thread got woken up ...
md0: no spare disk to reconstruct array! -- continuing in degraded mode
md: recovery thread finished ...
hdf: dma_timer_expiry: dma status == 0x61
hdf: error waiting for DMA
hdf: dma timeout retry: status=0xd0 { Busy }

hde: DMA disabled
hdf: DMA disabled
PDC202XX: Primary channel reset.
PDC202XX: Secondary channel reset.
ide2: reset: master: error (0x00?)
blk: queue cc826544, I/O limit 4095Mb (mask 0xffffffff)
md: ide/host2/bus0/target0/lun0/disc [events: 00000003]<6>(write) ide/host2/bus0/target0/lun0/disc's sb offset: 312571136
hdf: dma_timer_expiry: dma status == 0x41
hdf: error waiting for DMA
hdf: dma timeout retry: status=0xd0 { Busy }

hdf: DMA disabled
PDC202XX: Primary channel reset.
PDC202XX: Secondary channel reset.
ide2: reset: master: error (0x00?)
end_request: I/O error, dev 21:40 (hdf), sector 312571904
raid5: Disk failure on ide/host2/bus0/target1/lun0/disc, disabling device. Operation continuing on 1 devices
end_request: I/O error, dev 21:40 (hdf), sector 312572153
end_request: I/O error, dev 21:40 (hdf), sector 312572154
end_request: I/O error, dev 21:40 (hdf), sector 312572155
**snip**
end_request: I/O error, dev 21:40 (hdf), sector 312572157
end_request: I/O error, dev 21:40 (hdf), sector 312572158
end_request: I/O error, dev 21:40 (hdf), sector 312572159
md: updating md0 RAID superblock on device
md: (skipping faulty ide/host2/bus1/target1/lun0/disc )
md: (skipping faulty ide/host2/bus0/target1/lun0/disc )
md: ide/host2/bus0/target0/lun0/disc [events: 00000004]<6>(write) ide/host2/bus0/target0/lun0/disc's sb offset: 312571136
md: recovery thread got woken up ...
md0: no spare disk to reconstruct array! -- continuing in degraded mode
md: recovery thread finished ...

Electro · 07-15-2006, 03:45 PM

Promise controllers suck in Linux. The kernel developers added a lot of work arounds to get Promise controllers to work and they still have problems. I recommend using Highpoint Rocket133 or buy a 3ware card. I do not recommend setting up RAID-5 if it is a single processor system. RAID-5 consumes a lot of resources, so I recommend at least a two processor system. It is better to place one drive per IDE controller instead of three on one IDE controller. Do not use Seagate and Maxtor IDE or SATA drives for long term use like in a RAID array.

I do not know if the file server is for home or business. You may want to set up multiple RAID-1 and add them to LVM or EVMS. When you need more space, you can add more drives. You still get reduntancy but you also get infinite space.