[ntp:questions] NTP stops responding

A C agcarver+ntp at acarver.net
Mon Aug 22 19:55:34 UTC 2011


On 8/22/2011 07:38, A C wrote:
> On 8/21/2011 23:42, David Woolley wrote:
>  > A C wrote:
>  >> I've got a separate question from the PPS issue. Has anyone
>  >> experienced a situation where ntpd stops responding completely?
>  >
>  > I've never heard of that happening. However, ntpd will stop running if
>  > it appears that the local time is off by more than 1000 seconds. This
>  > can be disabled for the first setting.
>
> The time was already within a few milliseconds of the available network
> clocks but I was using a GPS receiver as one of the available clocks.
> However, the PPS isn't working (see the other thread for that issue) so
> the log file was full of "kern PPS no signal" messages and eventually
> stopped doing anything. Logging stopped, responses to queries stopped,
> everything except an explicit kill.
>
> I removed the GPS refclock completely and so far it's fine. I'll see how
> it looks tomorrow but it seems like the two issues may be related.

I've let ntpd run twice as long as it did yesterday and there's no sign 
of it hanging.  Therefore it appears that ntpd will hang when the GPS 
NMEA refclock is configured, kernel PPS (flag3) is enabled, and the PPS 
from the kernel isn't available for an extended period of time (this 
goes right back to my ongoing issue with PPS and ntpd).

I'm going to leave flag3 off and make sure the refclock isn't blowing up 
on its own.  I still want to get PPS working somehow but I just need to 
make sure ntpd is going to stay running.



More information about the questions mailing list