topic System Administration in Operating System - HP-UX

System Administration

MARIE ELDRIDGE — Tue, 27 Jun 2000 14:07:07 GMT

Is there an easy way to locate and maintain system and application logs so you don't run out of disk space? I would like to run a cron job to perform this maintenance.

Re: System Administration

Andy Monks — Tue, 27 Jun 2000 14:06:01 GMT

'sam' as an option do to this. just does some of them however.

Re: System Administration

Rick Garland — Tue, 27 Jun 2000 14:22:35 GMT

There are many tools out there for this purpose.
SAM is one good tool to keep on top of this. Also in the porting archive
is a trimlog binary. You can do the configs for what is to be done and all that good stuff. You could also write a script to do the tasks. Be careful though. some of the log files will need to be handled with care. Example is the syslog.log file. You may want to cp it somewhere then cat /dev/null > syslog.log because it is an open file. You may want to kill -HUP `cat /var/run/syslog.pid` to have it restart as well.

Re: System Administration

CHRIS_ANORUO — Tue, 27 Jun 2000 14:29:01 GMT

YES for example, put this line in your cron file:
0 08 27 12 * cat /dev/null > /var/adm/wtmp

Cheers!

Re: System Administration

James Odak — Tue, 27 Jun 2000 18:27:55 GMT

Assuming all logs end in .log or .LOG
write a script like this and add it to your cron

find / -depth -print|grep -i .log$ > /tmp/loglist

for X in `cat /tmp/loglist`
do
tail -50 $X > /tmp/templog
rm $X
mv /tmp/templog $X
done

This is the most simple way
you should add an if statement making sure /tmp/templog exists before removing and copying over the log file

you can also add -size +2000 to the find
this will limit the truncating of log files to just large ones

add a line with in the for statement

echo $X >> somefile

to keep a log of what files were clipped

you can also not delete but rename you log files each day

mv $X $X`date "+%m%d%y"`.log
then after the for statement run another find command
find / -depth -print -atime +x -exec rm {} /;

x being the number of days old the file deleted should be

i am sure there are more in depth ways to do this for people who script much better than i

and remember use sam for system files its better and much eaiser

Re: System Administration

James Odak — Tue, 27 Jun 2000 18:32:13 GMT

um when i put in find / blah blah
i should of said /dir of the dir your application is in
if you do just / you will get the.log of the system files that sam would take care of for you

sorry

Jim

Re: System Administration

MARIE ELDRIDGE — Fri, 14 Jul 2000 16:28:39 GMT

Thanks everyone for your help. Sorry I am late in responding, but I was out sick for a while. I never knew I had so many log files to keep track of.

Thanks again!

Re: System Administration

Alan Riggs — Fri, 14 Jul 2000 17:56:00 GMT

One thing to watch for in the scripted example: some logs have open file pointers at all times, which means that removing the file and moving another into place will corrupt the pointer reference (meaning you lose logging until the appropriate process is recycled). A safer way is:

tail -50 $LOGFILE > $TMPFILE
cat $TMPFILE > $LOGFILE

Note: tailing from the logical EOF is dangerous if you really want to preserve a large chunk of information, since the buffer will only hold 20K. That shouldn't be a worry for this problem, though.

Re: System Administration

Bill Hassell — Fri, 14 Jul 2000 23:41:10 GMT

There are two script samples: dailylogs.sh and weeklylogs.sh located at:

ftp://contrib:9unsupp8@hprc.external.hp.com/sysadmin/cronjobs

The dailylogs script will trim the fast growing files, where weeklylogs will trim a given logfile if it exceeds a set value. Each script is designed to be expanded and modified for local situations.