On 2008-05-08, Chris Kuethe <[EMAIL PROTECTED]> wrote:
> On Thu, May 8, 2008 at 8:52 AM, Brian A. Seklecki
><[EMAIL PROTECTED]> wrote:
>> Nagios checks almost never have sufficient debugging mechanisms, and UDP
>> services dont send RST+ICMP.
>
> you should get an ICMP port unreachable if there is no UDP service listening.
>
> i haven't looked at nagios, but i wonder if it's not trying to use NTP
> "mode 6" control messages to get more status information out of the
> daemon. openntpd doesn't support these queries...
check_ntp_peer does exactly that.
>> You an always: $ sudo ntpdate -qdv [host to check]
>
> or "rdate -pnv <host>". quite some time ago i added a check to make
> rdate bail out if the server is unsync'd.
>
> ...
> if ((data.status & STATUS_ALARM) == STATUS_ALARM) {
> warnx("Ignoring NTP server with alarm flag set");
> return (-1);
> }
> ...
>
> CK
>
check_ntp_time says "NTP CRITICAL: Offset unknown|" if that happens,
same as if the server isn't running. Not quite as much information as it
could give, but if you're basically looking to be alerted when your server
is broken, it's still helpful.
