On Tuesday 17 July 2007 17:45, Dustin Puryear wrote:
That's a FINE question, and there are a million ways to do this. Two quick ideas are:
- Use something like swatch or logwatch against your replication log
files and syslog (local4 if I remember correctly) logs.
- Use Big Brother (insert preferred monitoring software <here>), write a
simple log parser agent, and then trigger an alert if there is ever unexpected or error output in the log.
I prefer exception-style reporting, so the second option works better for me.
Thanks. Following your advice I found a perl script called slurpd_status.pl (*) that I integrated to my nagios server. It monitors entries in transition, rejected and waiting. I did a few satisfactory tests. What else do you think I should do concerning replication? I know slurpd is deprecated and I plan to move to syncrepl ASAP.
I'm using logcheck (from the FreeBSD ports) to monitor my system logs. AFAICS it is similar to logwatch and can certainly be used to monitor slapd's logs. What unexpected or error output do you think should raise alarms? What loglevel do you recommand for slapd?
Regards, Thierry.