Hey,

I have a CentOS running on a remote web server. It has a 10G HDD storage space and now it's almost full.
The thing is a while back i've accidentally done a 'crontab -r' instead of 'crontab -e' so all the log rotation/cleanup scripts are gone.

This being said, the logs for the domains keep growing bigger so I decided to delete the logs manually. I found the logs in /var/www/vhosts/domain.com/statistics/logs and i've deleted them. About 2G of logs. I used 'rm -rf'.

The problem is that the logs no longer appear when i do a 'ls -al' but space on disk isn't cleared either.

Do you guys have any idea ?

Thanks.

Your space didn't come back because those files are still open by the processes that are writing to them.

So, restarting those processes is necessary to get back your space.

A reboot of the server is the easiest way to accomplish this. But, if you can't do it for uptime purposes, then you'll need to look through /etc/rcx.d (for whatever run level you're running at) see if you can spot the program(s) that are writing to those files, and see if you can re-run them with a "restart" or "reload" command. Or, perhaps just a "stop" and then a "start" command.

For your future information, if you ever need to delete the contents of a log file in the future that you know is still open ( you'd either know it from past experience or you can see it from "lsof") without actually deleting the file itself, you can use the following command.

$> :> logfile.log

The ":" is a null, which makes the above command put a "null" in "logfile..log"

You could do this in the future to "clean" the file, without having to shutdown the programs. Of course a better solution is to use one of the rotatelog programs, but you'd probably have to shutdown the service writing to the file first, rotate the logs, and restart the service.

i more or less had the same problem. thank you for the answer!

Thanks a lot.

I only had to restart ( via 'service'): httpd and mysqld and ~2.5G cleared up. Thanks for the 'null' tip, i'll use it later on.

If my log rotate crontab is gone, can i create some scripts to put null into my logs form time to time ? That should be ok, should it not ?

Thanks again.

Sure, you can do that. But, you probably want some kind of way to keep and protect those log files (if this is a serious IT shop's application), at least for a while.

What I'd probably do is have a cron job that every day copies the contents of that file to some place else, an archive location.
Then, null the file,
Then exit

I've not even run it, but something like the following:

---WARNING UNTESTED - FUNCTIONAL PROTOTYPE ONLY ---

#!/bin/bash
$MYLOGDIR=/mydir
$MYARCHDIR=/myarchivedir
$MYLOGFILE=mylogfile.log
$FULL_LOG=$MYLOGDIR/$MYLOGFILE
if [ -f $FULL_LOG ]; then
cp $FULL_LOG $MYARCHDIR/$MYLOGFILE.`date "+%y%m%d"`
if [ $? -ne 0 ]; then
echo error copying file $FULL_LOG to $MYARCHDIR
exit 1
fi
:> $FULL_LOG || ( echo Error nulling file $FULL_LOG; exit 3)
echo "Archive and truncing $FULL_LOG complete"
else
echo "Archiving not needed: file $FULL_LOG nonexistent"
fi