UNIX for Dummies Questions & Answers

# see if the file exists
if [ -e full.log ]
then
# check the number of lines
cur_log_line=$(cat full.log | wc -l)
# compare to entry in saved file
lst_log_line=$(cat lst_log_line_file)
if [ cur_log_line -gt lst_log_line ]
then
# get lines after the last pass-thru
tail_lines = cur_log_line - lst_log_line
tail -$tail_line full.log > today.log
# set the last line checked
echo $cur_log_line > lst_log_line_file
# mail the file
uuencode today.log today.log | mailx -s "logs" somebody@me.com
fi
fi

p.s. funny how this board removes leading spaces - the code was more readable when indented

On Solaris 10 you can use the "logadm" utility to make any file rotate.

So configure logadm is such a way that a specific file is rotated at fixed times.

Suppose the name of this file is X.
After rotation the name of this file will be X.0 and there will be a new empty file X.

Create cron job which executes a few minutes after rotation of the log file has taken place, and which verifies the file X.0 for the presence of the messages you are interested in and takes the necessary actions e.g. emailing the result.

Thanks all for the help.
The particular log file is an Oracle log file for a particular process they do.
I'll give this a try and post any questions here. Thanks again!

I want to check the file for a particular line of text between checks. I was going to have a cron job rerun periodically and only check any new additions to the file. I can't mv the file as Oracle keeps it open.

That's probably the last piece of the puzzle. Thanks all!

mv /var/tmp/copy1 /var/tmp/copy2 2> /dev/null
cp <oracle log file> /var/tmp/copy1

diff /var/tmp/copy1 /var/tmp/copy2 | grep -v ^[0-9] | sed -e "s/^[<>] //" | grep <search string>