scsi0 : resetting for second half of retries

[Petr Her~man]

Pokousim se respektovat naznacena pravidla konference a psat az po jednom
dnu ruzneho vlastniho snazeni i cteni manu apod.

Vcera rano jsem prisel do prace a byl 'spadly' Linuxovy NFS & Samba
server - neslo se zalogovat a vubec nic. Po rebootu se jadro dostalo az k:

...
Linux version 1.2.13 ....
Partition check:
	sda: sda1 sda2 sda3
	sdb: sdb1 sdb2 sdb3 sdb4
	hda: hda1 hda2
	hdb: hdb1
	hdc: hdc1
VFS: Mounted root (ext2 filesystem) readonly.
scsi0 : reseting for second half of retries.
----------------------------------------------------
... a tady to konci.

Po ruznych neuspesnych pokusech s hybanim desek, kabelu atd. (predpokladal
jsem nejaky HW problem kolem SCSI) jsem nakonec na partition sdb2
(puvodne swap) nainstaloval base system Debian 1.1 (ktery ma tu vyhodnou
vlastnost, ze se v takovychto pripadech da do maleho prostoru snadno
nainstalovat) a ten pak z disketty (protoze z sdb2 se neda bootovat)
nabootoval. Behem bootovani jadra jsem dostal:

...
Checking file systems...
Parallelizing fsck ver. 1.02
/dev/sdb2: clean
scsi0: MEDIUM ERROR on channel 0, id 0, lun 0, CDB: 0x03, 00 00 00 10 00
Current error sr08:03: sns = f0  3
ASC=11  ASCQ= 0
Raw sense data: 0xf0 0x00 0x03 0x00 0x0b 0x9a ... atd.
scsidisk I/O error: dev08:03, sector 623342
Error reading block 311671 (Attempt to read block from filesystem resulted
in short read) while reading indirect blocks of inode 77842

/dev/sda3: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.

fsck failed. Please repair manually and reboot. ... atd.

Z toho jsem usoudil, ze je nejaka chybka na /dev/sdb3 coz je puvodni
rootova partisna. Takze po rootovskem heslu:

# mount -n -o remount,rw /
# fsck /dev/sda3
Pass 1:
scsi0: MEDIUM ERROR on channel 0, id 0, lun 0, CDB: 0x03, 00 00 00 10 00
Current error sr08:03: sns = f0  3
ASC=11  ASCQ= 0
Raw sense data:0xf0 0x00 0x03 0x00 0x0b 0x9a 0xee 0x0e 0x00 0x00 0x00 0x00 0x11
0x00 0x00 0x80
scsidisk I/O error: dev08:03, sector 623342

Error reading block 311671 (Attempt to read block from filesystem resulted in
short read) while reading indirect blocks of inode 77842. Ignore error<y>? no

Inode 77842, i_blocks wrong 1108 (counted=596). Set i_blocks to counted<y>? yes

scsi0: MEDIUM ERROR on channel 0, id 0, lun 0, CDB: 0x03 00 00 00 10 00
Current error sr08:03: sns = f0  3
ASC=14  ASCQ= 0
Raw sense data:0xf0 0x00 0x03 0x00 0x0c 0xb1 0xd6 0x0e 0x00 0x00 0x00 0x00 0x14
0x00 0x00 0x80
scsidisk I/O error: dev08:03, sector 694742
Error reading block 347371 (Attempt to read block from filesystem resulted in
short read) while reading indirect blocks of inode 84163. Ignore error<y>? no

fsck.ext2: Attempt to read block from filesystem resulted in short read while
calling ext2fs_block_iterate in check_blocks

Inode 84163, i_blocks wrong 34 (counted=26). Set i_blocks to counted<y>? yes

Pass 2: checking directory structure

scsi0: MEDIUM ERROR on channel 0, id 0, lun 0, CDB: ...
Current error ...
ASC=...
Raw sense data: ...

... atd. a tady mi obrazovky jen roluji pred ocima. Nakonec:

Fix summary information<y>? yes

# shutdown -r now

Predpokladam, ze se to uz na tom disku opravilo, tak nabootuju znova puvodni
system ... no a presne to samy, jako na zacatku:

VFS: Mounted root (ext2 filesystem) readonly.
scsi0 : reseting for second half of retries.

Tak zkusim zase nabootovat jadro 2.0.6 z Debianovske disketty a dat puvodni
root:
boot: linux root=/dev/sda3
ale zase:

scsi0: MEDIUM ERROR on channel 0, id 0, lun 0, CDB: ...
Current error ...
ASC=...
Raw sense data...
scsidisk I/O error: dev 08:03, sector 623342
... roluji mi obrazovky, nakonec:
init: can't resolve symbol '__ctype_b'
... a konec.

Vypada to, jako by se to tim fsck vubec nespravilo (ten sektor je ten
samy), ale kdyz znova
nabootuju Debiana s root=/dev/sdb2 a pustim
#fsck /dev/sda3
tak mi rekne, ze je clean.
Ma prosim nekdo predstavu, co se to deje?

Diky
	Petr Herman