Hi All,

I need a short script that will be run from a cron.
It needs to grep the size of a file.
If the file size is bigger than 500MB, Delete that file.


Any pointers?

thanks,

-Sup.

'man find' ?

I think the problem here is not to find files bigger than a certain size, but to see if a particular file is grown up to that size, right? The find command would be very inefficient if the file resides in a upper level of a directory tree. So, why not simply something like...?

I have the file name. It is a log file that every other weeks is going crazy due to a process running uncontrolled.

I need something like:

This script will then be run via cron every 10 minutes.

I just need the right syntax for the code.

thanks,

-Sup.

Except man find is really hard to work though.

Try something like

find /path/to/start/looking/from -size +500M -exec rm {} \;

Try it without the part from -exec first, just to make sure it's showing the files you actually want to delete (ie "find /path/to/start/looking/from -size +500M")

Then read "man find".

Edit - you may want to change that to "... rm -f {} ..." depending on the user.

I wrote my bit of the code the same time you wrote yours ) so I saw your post after I replied.

Yes, I think this will work out great!

test $(stat -c \%s /var/log/httpd/error_log) -gt 500000000 && rm -Rf /var/log/httpd/error_log


The only addition I've added was the -Rf so 'don't ask me questions' when it runs.

I now just need to set it in a cron and Restart apache.

Thanks much!

-Sup.

Well, if it is a log file you have to stop the process to which the file belongs, before removing. Otherwise the file is not actually removed, that is the inode is preserved (due to the fact the file is in use) and the disk space is not freed.

Or just

find /var/log/http -maxdepth 1 -name 'error_log' -size +500M -exec rm -f {} \;

Edit - notwithstanding comment above.

Wouldn't logrotate be good for this too (not every 10 minutes perhaps)?

crontab -e
*/10 * * * * /etc/cleanApacheErrorLogFile.sh

Does this gets me closer to my target?

Thanks,

-Sup.

Your solution also sounds good

logrotate is currently configured to run it's normal rotation for this log as well as for other logs within the same path. I don't think you can make logrotate run just for a single file in that path. Also logrotate runs on cron.daily, and I need this to run on 10 minutes intervals.

Thanks,

-Sup.

Hi All,

I think I'm not able to run the process correctly and need your input on this:

I need the process to delete the error_log file once it is > 500MB.
Just before it deletes it, it needs to STOP httpd, Delete the file, then Restart httpd.

I tried to run this in a cron:


server.myserver.com:/etc # crontab -e

*/10 * * * * /usr/bin/rdate time-a.nist.gov >date -s
0 5 * * 0 /usr/local/sysbk/sysbk -q
*/10 * * * * /etc/cleanApacheErrorLogFile.sh


server.myserver.com:/etc # less cleanApacheErrorLogFile.sh
test $(stat -c \%s /var/log/httpd/error_log) -gt 500000000 && service httpd stop && rm -Rf /var/log/httpd/error_log && service httpd
restart


But, the script doesn't want to run when in crontab.

If I ./cleanApacheErrorLogFile.sh then it runs fine and does what it should.
But when I try to let it run by crond, it doesn't run.


Any pointers on how to make this work correctly?

Thanks,

-Sup.

Cron has a very limited environment, for example the PATH is usually /bin:/usr/bin. Therefore it is a good rule to use full path for every command in your script. Most likely the service command, which is usually in /sbin, is not found. Check also the mail of root: all the standard output and the standard error which is not explicitly redirected to a file is sent to the owner of the crontab (use the mail command to find out).

And so,. this is with explicit calls, and still nothing.
I'm confused.
As far as the mail tracing:

less /var/log/exim/mainlog


Any pointers on how to make this more usefull?

Thanks,

-Sup.

Finaly the solution that works and added some logging.

The solution was provided by tillo in the DirectAdmin support forums:

http://www.directadmin.com/forum/sho...d=1#post144619

###########

Placed in the /etc/cron.d/error_log


MAILTO=email@mydomain.com
* * * * * root find /var/log/httpd/error_log -size +500M -exec rm '{}' \; -exec /usr/sbin/apachectl restart \; -exec sh -c 'echo $(date): $1 erased and httpd restarted |tee -a /var/log/httpd/CleanApacheRestarts.log' {} \; 2>/dev/null


This seems to work fine, also adds the email of the event with timestamp and logs it on the server.

Thanks for everyone's help.


-Sup.