lkml.org 
[lkml]   [1997]   [Oct]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subjectrepeating error messages, printk behaviour, 2.0.31 problem
Hello,

we are seeing the following on a news server:
Oct 26 15:35:36 praxis kernel: EXT2-fs error (device 08:12): ext2_add_entry: bad entry in directory #408806: inode out of bounds - offset=1180, inode=268844812, rec_len=12, name_len=4
Oct 26 15:35:36 praxis kernel: EXT2-fs error (device 08:12): ext2_add_entry: bad entry in directory #408806: inode out of bounds - offset=1180, inode=268844812, rec_len=12, name_len=4
Oct 27 23:52:03 praxis kernel: EXT2-fs error (device 08:05): ext2_free_blocks: Freeing blocks not in datazone - block = 268601838, count = 1 Oct 29 21:05:13
praxis kernel: EXT2-fs error (device 08:12): ext2_readdir: bad entry in directory #474213: inode out of bounds - offset=156, inode=268911077, rec_len=12, name_len=4
Oct 29 21:43:30 praxis kernel: EXT2-fs error (device 08:12): ext2_readdir: bad entry in directory #474213: inode out of bounds - offset=156, inode=268911077, rec_len=12, name_len=4

The messages repeated about 17000 times per minute, making the poor
machine more or less unusable (the log messages caused about 550 context
switches [any -> klogd -> syslogd -> anything else] per second, and i
think they added a lot to the problems, as the machine is heavily loaded
anyway).
It was much worse when there were errors on both news drives at the same
time.

Maybe printk should filter out lines which repeat too often?



I'm pretty sure the actual errors are hardware related, but for the
records:

Software: 2.0.31 (did happen with 2.0.31-pre2, too). innd. satcomd
(satellite news feed).
Hardware: Pentium, 128MB Ram (64 at the moment), Adaptec AHA-294X,
3Com 3c900 Boomerang.
sda: Vendor: QUANTUM Model: FIREBALL_TM3200S Rev: 300N swap, system
sdb: Vendor: IBM Model: DORS-32160 Rev: WA6A news
sdc: Vendor: IBM Model: DCAS-34330 Rev: S61A news/alt
I didn't find any SCSI error messages in the log.
Unmounting/fsck/remounting the news partitions helps, the problems
vanish for some time (maybe 6 to 7 hours).

Regards, Uwe

\
 
 \ /
  Last update: 2005-03-22 13:40    [W:0.024 / U:0.172 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site