We recently started using Nagios to monitor a number of computer systems, as well some key services running on them. One such service is NTP, the Network Time synchronisation service (supported in Nagios via the check_ntp plugin).
Earlier today, Nagios (2.10) beganÂ to report that a number of our (Solaris) systems had drifted too far out of sync from their associated time servers. However, when we checked the time and NTP service on each system, we could not find anything wrong. To cut a long story short, it turns out that the system on which Nagios was running, was not itself synchronised to any form of time server and as such, it was the system whose time had actually drifted out of sync.
This is rather curious behaviour on the part of Nagios in my opinion and I wonder if it isn’t actually a bug in the check_ntp plugin. Why should it matter what the time is where Nagios runs so long as the systems being monitored haven’t drifted off their respective mark? Also, what it we were monitoring systems in different time zones?
As far a the solution to the original problem goes, this was quite straightforward in the end. All we had to do was configure our Nagios system to synchronise to the same time server as the systems it was monitoring. However, I still don’t believe we should have to do this.