John's suggestion is still pertinent. You'll need a SIGHUP handler in your script. Logrotate could send the SIGHUP in a postrotate 'script'. Cheers, Cliff On Sun, Dec 22, 2013 at 3:15 PM, Larry Martell <larry.martell at gmail.com>wrote: > On Sat, Dec 21, 2013 at 8:52 PM, John R Pierce <pierce at hogranch.com> > wrote: > > On 12/21/2013 4:56 PM, Larry Martell wrote: > >> I'm looking for advice or suggestions for rolling log files with a > >> daemon. I have a python script that I daemonized with > >> > http://www.jejik.com/articles/2007/02/a_simple_unix_linux_daemon_in_python/ > . > >> Before I daemonized it it was run from a bash script that invoked the > >> underlying python script. It ran the python script, waited for it to > >> complete and then it slept for 5 seconds and ran it again. This was in > >> a infinite loop. In between each invocation it checked the log file > >> and if it was over 10MB it renamed it and then the next invocation > >> started with a new empty log. Since each invocation was a separate run > >> this worked fine. But now the daemonized python script doesn't exit - > >> the same log file is attached to it forever. So my renaming of the > >> file does nothing - the i node doesn't change and it's still logging > >> to the same large file. Anyone have any ideas how I can achieve this > >> sort of log rolling in this situation? > >> > > > > send a SIGHUP to syslog and it shoudl re-opent he log files. > > > > silly question, but whats wrong with the logrotate daemon thats built > > into centos? > > This is not using syslog. If you look at the daemonizing script I gave > the link to, you pass in the log files for stdout and stderr, and it > does some double fork magic and then associates the given files with > them. > _______________________________________________ > CentOS mailing list > CentOS at centos.org > http://lists.centos.org/mailman/listinfo/centos >