The nagios event scheduler can fire off events somewhat early (up to 100ms), if an fd is touched during that time, to save some sleeps. This isn't generally a problem, with exception of log handling. But it doesn't save that much either.
We have a patch that removes those early timeouts, and makes an event guaranteed to fire of first when the event should be triggered:
http://git.op5.org/git/?p=op5-monitor-c ... eefe39fbb0
And then, in the case, by some magic, an event is fired of early, this patch makes the log file not beeing overwritten, but appended .1, .2, .3 and so on:
http://git.op5.org/git/?p=op5-monitor-c ... 83ec8fd2a6
Log rotates several times in a row, replacing old log files
-
scottwilkerson
- DevOps Engineer
- Posts: 19396
- Joined: Tue Nov 15, 2011 3:11 pm
- Location: Nagios Enterprises
- Contact: