Re: Why syslog is not rotating?

[Reco <recoverym4n@gmail.com>]

On Sun, 3 Nov 2013 17:16:02 +0200 (IST)
Itay <debian@itayf.fastmail.fm> wrote:

> On Sun, 3 Nov 2013, Reco wrote:
> > Now that's interesting. Is there anything similar to this messages
> > in /var/log/cron.log?
> 
> Hmmmm... there is no /var/log/cron.log !!

Sorry, my mistake. I have an old installation, /var/log/cron.log is a
leftover of etch's setup in my case. Do you have any MTA
(exim, sendmail or postfix) installed? Anytime cron job puts anything
to the stderr cron should send mail to the local root (default
settings). Is there anything suspicious in the root mailbox?
And, is there anything unusual in /var/log/kern.log at the time you
had this error?


> > Does, say, 'md5sum /var/log/syslog' runs to the completion?
> 
> Yes.  Without warnings/errors.
> 
> > What about 'cat /var/log/syslog > /dev/null'?
> 
> Yes.  Without warnings/errors.

Ok. What about 'cat /var/log/syslog | gzip -c > /dev/null'?
And, while we're at that, what about: 

cat /var/log/syslog | gzip -c > /var/log/syslog.test.gz

If error shows early, can you also post contents of (/tmp/gzip):

strace -fo /tmp/gzip cat /var/log/syslog | gzip -c > /dev/null


> > Can you run fsck on the filesystem containing /var/log/syslog?
> 
> I have to unmount /var for that; right?

Yes.


> So I need to use Live CD for that; right?

Sure, that's possible to do with livecd. But, you can also do it from a
single-user (i.e. init 1; unmount /var; run fsck on a logical volume;
reboot).


> > What does smartctl --all shows on the partition with this filesystem?
> 
> I never used smartctl (installed it now following-up your question).
> In my system /var resides on a logical volume.
> So I am not sure how to proceed.

Find a physical volume corresponding to the /var logical volume.
Run smartctl --all on the disk that's containing that physical volume.
In case you have RAID (be it mdadm or dm-mirror) - run smartctl on all
disks that are part of said RAID.

While we're on it, also run smartctl -t long on said disk, wait for a
while (smartctl should say you, how much), and run smartctl --all on
the same disk again.

Reco