[Nagiosplug-help] check_ntp

[Richard Mohr]

On Sat, 2007-11-10 at 00:10 -0500, Thomas Guyot-Sionnest wrote:

> check_ntp gets the offset between the local clock the the peer, which is
> good to check if the local server is sync with the remote server, but
> not to check if the remote server is sync with its peers.

That wasn't clear to me at first, and I suppose most of the fault lies
with me.  The C version I am using has this help message:

  This plugin checks the selected ntp server

That doesn't explain too much, so I just assumed that it was doing the
equivalent of "ntpq" and checking the offset, jitter, etc. that the
remote host had with its peers.

The perl version I am using has this help message:

  Checks the local timestamp offset versus <host> with ntpdate
  Checks the jitter/dispersion of clock signal between <host> and its sys.peer with ntpq

I just didn't read the message closely enough to realize how it was computing the offset.

> This can easily be changed; the question is do we actually want that?
> (I'm personally all for it...)

I would definitely want it.  I can see how checking the offset between
the Nagios server and the remote host would be useful.  (It would help
ensure that the timestamps for Nagios events match the timestamps in the
log files in the remote host.)

However, if the time on my Nagios server is off, but all my other hosts
are correct, using this offset would generate a lot of false errors.  I
mainly care if the remote host's time closely matches its peers.

-- 
Rick Mohr
Systems Developer
Ohio Supercomputer Center