I just noticed on one of our servers this... Code: -rw-r--r-- 1 root root 4.2M Jul 24 13:02 error.log -rw-r--r-- 1 root root 227K Feb 12 00:31 error.log.1.gz Normally access log's are rotated daily but how is the error log rotation done? PS. How do you feel about bind-mounting log folders (or www folder?)
The error log gets rotated when it exceeds a certain size, I'm not quite sure what the actual size is, it must be in the cron_daily.php code somewhere.
Yesterday one of our customers error log grew to 13Gb because of exploit found in PHP code. As a result that cluster node became full causing problems on the VM. I'm trying to find out how to prevent such behavior in the future. Any ideas, any1?
Currently, the size is just checked once a night, probably the file grew to that size in less than 24 hours. One option would be to have a script that checks the log sizes more often.
