Dotaz: vadne bloky na scsi disku

[tl]

Dobry den, mam SW raid 1 na serveru (SuSE 9.2) s dvema SCSI disky FUJITSU MAM3184MP. Obcas se stava, ze jedna partisna z raidu vypadne, ale vzdy to spravila kombinace raidsetfaulty + raidhotremove + raidhotadd. Dnes po raidhotadd probehla recovery asi do 60%, pak zustala prislusna partisna faulty a syslog ohlasil (vetsitka a mensitka jsem kvuli xhtml nahradil hranatymi zavorkami): Feb 23 09:50:54 balrog-old kernel: md: bind[sdb3] Feb 23 09:50:54 balrog-old kernel: RAID1 conf printout: Feb 23 09:50:54 balrog-old kernel: --- wd:1 rd:2 Feb 23 09:50:54 balrog-old kernel: disk 0, wo:0, o:1, dev:sda3 Feb 23 09:50:54 balrog-old kernel: disk 1, wo:1, o:1, dev:sdb3 Feb 23 09:50:54 balrog-old kernel: md: syncing RAID array md0 Feb 23 09:50:54 balrog-old kernel: md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc. Feb 23 09:50:54 balrog-old kernel: md: using maximum available idle IO bandwith (but not more than 200000 KB/sec) for rec onstruction. Feb 23 09:50:54 balrog-old kernel: md: using 128k window, over a total of 13735488 blocks. Feb 23 09:57:29 balrog-old kernel: SCSI error : [0 0 1 0] return code = 0x8000002 Feb 23 09:57:29 balrog-old kernel: Info fld=0x17bc5b3, Deferred sdb: sense key Hardware Error Feb 23 09:57:29 balrog-old kernel: ASC= 3 ASCQ=80 Feb 23 09:57:29 balrog-old kernel: end_request: I/O error, dev sdb, sector 24890315 Feb 23 09:57:29 balrog-old kernel: raid1: Disk failure on sdb3, disabling device. Feb 23 09:57:29 balrog-old kernel: Operation continuing on 1 devices Feb 23 09:57:29 balrog-old kernel: md: md0: sync done. Feb 23 09:57:29 balrog-old kernel: RAID1 conf printout: Feb 23 09:57:29 balrog-old kernel: --- wd:1 rd:2 Feb 23 09:57:29 balrog-old kernel: disk 0, wo:0, o:1, dev:sda3 Feb 23 09:57:29 balrog-old kernel: disk 1, wo:1, o:0, dev:sdb3 Feb 23 09:57:29 balrog-old kernel: RAID1 conf printout: Feb 23 09:57:29 balrog-old kernel: --- wd:1 rd:2 Feb 23 09:57:29 balrog-old kernel: disk 0, wo:0, o:1, dev:sda3 Partisnu jsem vyhodil z raidu, pustil na ni badblocks (badblocks -s -o badblocks_sdb3 -v /dev/sdb3) a zformatoval ji s oznacenim nalezenych vadnych bloku (reiserfsck -B badblocks_sdb3 --rebuild-tree /dev/sdb3). Reiserfsck dvakrat zhavaroval na dalsich vadnych blocich ("bread: Cannot read the block (2016599): (Input/output error)."), po jejich pridani do souboru s cisly vadnych bloku uz probehl normalne. Partisna pak sla pripojit, slo z ni cist i do ni zapsat, ale kdyz jsem se ji pokousel pridat do raidu, raidhotadd skoncil zase na cca 60% stejnou hlaskou. Kdyz jsem pak partisnu znovu formatoval, reiserfsck zhavaroval na dalsim vadnem bloku a v syslogu se objevilo: Feb 23 10:03:44 balrog-old kernel: scsi0: ERROR on channel 0, id 1, lun 0, CDB: Read (10) 00 01 7b c5 0b 00 01 00 00 Feb 23 10:03:44 balrog-old kernel: Info fld=0x17bc5b7, Current sdb: sense key Medium Error Feb 23 10:03:44 balrog-old kernel: Additional sense: Read retries exhausted Feb 23 10:03:44 balrog-old kernel: end_request: I/O error, dev sdb, sector 24888758 Feb 23 10:03:44 balrog-old kernel: Buffer I/O error on device sdb3, logical block 8243381 Feb 23 10:03:45 balrog-old kernel: Buffer I/O error on device sdb3, logical block 8243382 Oba ty uvedene logicke bloky oznacil badblocks hned na zacatku. Projel jsem to cele jeste jednou, vysledek byl stejny. Je ten disk na vyhozeni (vadna partisna na nem zabira cca 3/4) nebo se da jeste neco vyzkouset (nahradni disk momentalne nemam a server nemuzu odstavit)?

Odpovědi

[Ondřej Čečák]

Proc je ten dotaz tak zpraseny do jednoho odstavce? :)

Doporucuji pouzit S.M.A.R.T. a podivat se, jak to je s chybami, co o sobe ten disk rika a zkousit nejaky test -- samozrejme nejlepe na jinem stroji a po te disk poslat na reklamaci nebo do popelnice.