Topic: SCSI error on additional raid5

[fpausp]

Hi All,

I try to find out a problem on my sme 7.4, i installed an additional raid5 with the following command:

Code: [Select]

mdadm --create /dev/md3 --level=5 --raid-devices=6 /dev/sda1 /dev/sdb1 /dev/sdc1 /dev/sdd1 /dev/sde1 /dev/sdf1

I did a quick search in /var/log/messages with:

Code: [Select]

cat -n /var/log/messages | grep "SCSI error"

193  Feb 23 12:21:59 djfjfd kernel: SCSI error : <0 0 2 0> return code = 0x8000002
199  Feb 23 12:21:59 djfjfd kernel: SCSI error : <0 0 1 0> return code = 0x8000002


sed -n '193,204p' /var/log/messages

Feb 23 12:21:59 djfjfd kernel: SCSI error : <0 0 2 0> return code = 0x8000002
Feb 23 12:21:59 djfjfd kernel: Info fld=0x0, Current sdc: sense key Hardware Error
Feb 23 12:21:59 djfjfd kernel: Additional sense: Internal target failure
Feb 23 12:21:59 djfjfd kernel: end_request: I/O error, dev sdc, sector 1953150399
Feb 23 12:21:59 djfjfd kernel: md: write_disk_sb failed for device sdc1
Feb 23 12:21:59 djfjfd kernel: raid5: Disk failure on sdc1, disabling device. Operation continuing on 5 devices
Feb 23 12:21:59 djfjfd kernel: SCSI error : <0 0 1 0> return code = 0x8000002
Feb 23 12:21:59 djfjfd kernel: Info fld=0x0, Current sdb: sense key Hardware Error
Feb 23 12:21:59 djfjfd kernel: Additional sense: Internal target failure
Feb 23 12:21:59 djfjfd kernel: end_request: I/O error, dev sdb, sector 1953150399
Feb 23 12:21:59 djfjfd kernel: md: write_disk_sb failed for device sdb1
Feb 23 12:21:59 djfjfd kernel: raid5: Disk failure on sdb1, disabling device. Operation continuing on 4 devices


Is it a faulty drive? A faulty SCSI card? Is it a Linux bug in the kernel or in some driver ?

[CharlieBrady]

Is it a faulty drive? A faulty SCSI card? Is it a Linux bug in the kernel or in some driver ?

Any of those, but faulty cable is probably more likely, followed by faulty drives (sdb and sdc).

[piran]

Do as Charlie advises but also consider checking the SCSI termination arrangements.

[fpausp]

OK, i will check the cables first ...

[smeghead]

T.. also make sure your PSU is up for the task of running all this extra kit.  If it's weak or undersized then the SCSI bus may not be able to initialise some of the drives properly.  An intelligent SCSI controller will often stagger the spin up of the drives to reduce PSU load, at least all my HP Smart Array controllers do, as do the Intel RAID cards I've used.

Try booting the server from cold & then straight away do a warm restart, with the drives already spinning there is less chance of this type of issue arising; now try the same routine & see if it gives a diff result.

If still no result post up the type of server, controller card, cable/connection type & drives used.  Someone may have info about certain combos (eg:  I've had issues with IBM scsi drives with a firmware version that was intemittantly flakey with Adaptec RAID controllers).

HTH

[fpausp]

Hi All,

Thanks for your professional reply, in the meantime my hardware-distributor made a test, he said the backplane is the problem.

In two weeks i will reinstall the server, we will see ...

Best
fpausp