Can you force a drive to red ball/become disabled?

[glave]

I'm in the midst of rebuilding a  drive in the array, and it appears one of my two cache drives has gone flaky. I'd like to disable that drive without needing to cancel my current rebuild. It's currently causing some very bad system slowdowns as it's trying to write  and timing out on this drive.

 

Dec 27 13:10:59 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3168, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:10:59 arcade kernel: sd 6:0:0:0: [sdm] tag#9 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x06
Dec 27 13:10:59 arcade kernel: sd 6:0:0:0: [sdm] tag#9 CDB: opcode=0x2a 2a 00 05 20 9e 40 00 05 00 00
Dec 27 13:10:59 arcade kernel: blk_update_request: I/O error, dev sdm, sector 86023744
Dec 27 13:47:01 arcade root: Parity Check / Rebuild Running - Not executing mover
Dec 27 13:58:27 arcade kernel: sd 6:0:0:0: [sdm] tag#1 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x06
Dec 27 13:58:27 arcade kernel: sd 6:0:0:0: [sdm] tag#1 CDB: opcode=0x2a 2a 00 09 07 13 40 00 05 00 00
Dec 27 13:58:27 arcade kernel: blk_update_request: I/O error, dev sdm, sector 151458624
Dec 27 13:58:27 arcade kernel: btrfs_dev_stat_print_on_error: 13 callbacks suppressed
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3182, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3183, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3184, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3185, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3186, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3187, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3188, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3189, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3190, rd 0, flush 0, corrupt 0, gen 0
Dec 27 13:58:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3191, rd 0, flush 0, corrupt 0, gen 0
Dec 27 14:14:27 arcade kernel: sd 6:0:0:0: [sdm] tag#17 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x06
Dec 27 14:14:27 arcade kernel: sd 6:0:0:0: [sdm] tag#17 CDB: opcode=0x2a 2a 00 0a b1 49 c0 00 05 00 00
Dec 27 14:14:27 arcade kernel: blk_update_request: I/O error, dev sdm, sector 179390912
Dec 27 14:14:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3192, rd 0, flush 0, corrupt 0, gen 0
Dec 27 14:14:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3193, rd 0, flush 0, corrupt 0, gen 0
Dec 27 14:14:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3194, rd 0, flush 0, corrupt 0, gen 0
Dec 27 14:14:27 arcade kernel: BTRFS error (device sdm1): bdev /dev/sdm1 errs: wr 3195, rd 0, flush 0, corrupt 0, gen 0

[trurl]

I would suggest you quit writing to your server while the rebuild is in progress. After a successful rebuild, then you can try to sort out your cache drive issue.

[garycase]

Ditto => the cache drive has nothing to do with a drive rebuild.  In fact even if you're writing to it there shouldn't be any impact on the rebuild -- are you sure you're not writing to the array?  [which I agree you should avoid during a rebuild]