Files corrupt - larger then supposed to be - redhat 9

[Geoff Wild]

Help!

Noticed cron hasn't been starting - get tis in boot.log:

Apr 22 13:06:56 dune crond: crond startup failed
Apr 22 13:06:57 dune xfs: xfs startup succeeded
Apr 22 13:06:57 dune anacron: execvp: File too large
Apr 22 13:06:57 dune anacron: anacron startup failed

So, I went to /usr/sbin and look at what an ls -al returns:

# ls -al anacron
-rwxr-xr-x 1 root root 2251804108673568 Jan 24 2003 anacron

How did that happen???

There a re a few others:

-rwxr-xr-x 129 8388608 8388736 74309398146588960 Dec 8 12:57 rotatelogs

-r-xr-xr-x 1 root 8388608 2251804108662572 Jan 25 2003 yppoll
-r-xr-xr-x 129 8388608 8388736 2251804108660196 Jan 25 2003 ypset
-r-xr-xr-x 129 8388608 128 74309398146593192 Jan 25 2003 yptest


When I try to restore :

Message from syslogd@dune at Thu Apr 22 12:29:40 2004 ...
dune kernel: Assertion failure in journal_forget_R89e31bd9() at transaction.c:1255: "!jh->b_committed_data"

Same if I try rm, or cat /dev/null > anacron

Please, tell me I can recover somehow without having to re-install.....

Rgds...Geoff

[Geoff Wild]

Okay - all fixed - did a:

shutdown -r -F now

which forced a fsck - and that cleared it up - had to be on the console though...


Rgds...Geoff

[Stuart Browne]

The 'anacron' and the 'crond' servies are different.

'anacron' is designed for people who don't have the machine on 24/7, allowing for people to turn machine on, and have out-standing scheduled jobs to run (so you can have the weekly log rotation, even if you only turn it on twice a week).

'crond' is still the same beast on most other unicies, being the core system scheduler (combined with 'at').

Both have different configuration files ('/etc/anacrontab' for anacron, '/etc/crontab'+'/var/spool/cron/'+'/etc/cron.d/' for crond)

[Steven E. Protter]

Goeff, I'd recommend talking a close look at the system to see why those nice filesystems got corrupt. There could have been an issue with mirrors or the disk.

I'd be very careful about backing up data on this box and watch ir really closely for signs.

Anything in /var/log/messages

SEP

[Geoff Wild]

I'm pretty sure it was me - the time it started was the time I was playing with logrotate - and adding awstats/virtual servers to the /etc/logrotate.f/httpd file - I must have done something with cut'n'paste or something or didn't have the output/redirect configure correctly - I checked the system for intusion, as well as other files - as well as disk errors - everything checks out fine.

Thanks...Geoff

[U.SivaKumar_2]

Hi,

Did you try to fsck ( forced ) that filesystem ?

regards,

U.SivaKumar.

[U.SivaKumar_2]

Hi,

I strongly suspect that your journal is corrupt. Recreate the journal.
Boot in single user mode.

#tune2fs -O ^has_journal /dev/sdX
#tune2fs -j /dev/sdX

regards,

U.SivaKumar.