Shell Programming and Scripting

Hi, I would like to know how to go about writing a script to compress & deleate old files from /var/mqm/log file system.

I am a complete beginner and would love it if someone could actually give me the code for this.

Thank you.

You want to compress & delete the same files? Then why compress them? Why not just delete them? Also, how old is "old"? Can you be more specific?

#!/usr/bin/ksh

LOGDIR=/var/mqm/log
COMPRESSDAYS=3
DELETEDAYS=5

find ${LOGDIR} ! -name \*.gz -mtime +{COMPRESSDAYS} -exec gzip {} \;
find ${LOGDIR} -name \*.gz -mtime +{DELETEDAYS} -exec rm {} \;


This script would compress all files in your log dir which are older than 3 days and delete all compressed files which are older than 5 days.

However, many Unix systems have a standard facility to arrange something like this for you, like "logadm" on Solaris systems.

Please suggest me what to do? I have to compress the log file and move it and then later delete it. Never worked on it please suggest the best way to do it. I have to run this job on corntab.

Some one wrote this for me please explain How this exactly works? Any help is appreciated.

#!/bin/ksh
DATE=`date +%Y%m%d%H%M`
FILES=`find /soa/p2/syst/var/log -name XMLTTrace.log`
for logfile in $FILES
do
newlogfile=$logfile.$DATE
CMD="cp $logfile $newlogfile"
$CMD
cat /dev/null > $logfile
rotated_logs="$rotated_logs $newlogfile"
done
for newlogfile in $rotated_logs
do
echo $newlogfile
echo "look in /soa/p2/syst/var/log/XMLTLogs for backed up log files"
echo "files gzipped. files deleted after one week"
/usr/bin/gzip -9 $newlogfile
mv ${newlogfile}.gz /soa/p2/syst/var/log/XMLTLogs
find /soa/p2/syst/var/log/XMLTLogs -mtime +7 | xargs rm
done

Apparently the directory "/soa/p2/syst/var/log" contains subdirectories.

In each of these subdirectories the file "XMLTTrace.log" can be present.

All the files named "XMLTTrace.log" located in the directory "/soa/p2/syst/var/log" or it's subdirectories are located using the find command.

For all those files found, the file is copied to a new name with a time stamp appended to it using the "cp" command.

Next the original file is truncated using the "cat" command


All files which are found are also stored into the variable "rotated_logs"


#!/bin/ksh
DATE=`date +%Y%m%d%H%M`
FILES=`find /soa/p2/syst/var/log -name XMLTTrace.log`
for logfile in $FILES
do
newlogfile=$logfile.$DATE
CMD="cp $logfile $newlogfile"
$CMD
cat /dev/null > $logfile
rotated_logs="$rotated_logs $newlogfile"
done

All file stored in the variable "rotated_logs" are next compressed and moved to a different directory; "/soa/p2/syst/var/log/XMLTLogs"

This directory is also searched for files older than 7 days, and those are removed.

for newlogfile in $rotated_logs
do
echo $newlogfile
echo "look in /soa/p2/syst/var/log/XMLTLogs for backed up log files"
echo "files gzipped. files deleted after one week"
/usr/bin/gzip -9 $newlogfile
mv ${newlogfile}.gz /soa/p2/syst/var/log/XMLTLogs
find /soa/p2/syst/var/log/XMLTLogs -mtime +7 | xargs rm
done

All in all, this is a very inefficient scripts. Just look at the last "find" command which is inside the "for ... do ... done" loop, meaning the same action is performed several times where only the first time the find runs can really do something.


All could be done in a single "loop".


#!/bin/ksh
DATE=`date +%Y%m%d%H%M`
find /soa/p2/syst/var/log -name XMLTTrace.log | \
while read FILE
do
mv ${FILE} ${FILE}.${DATE}
touch ${FILE}
gzip -9 ${FILE}.${DATE}
mv ${FILE}.${DATE}.gz /soa/p2/syst/var/log/XMLTLogs
done
find /soa/p2/syst/var/log/XMLTLogs -type f -mtime +7 -exec rm {} \;

This is still not the way I would do it, however it's a more efficient version of your script. The only "but" here is that the newly created log file could have different ownerships/permission as the original one because instead of being truncated it is newly created. This could solved by placing a "chown" and "chmod" command between the "touch" and "gzip" lines.