Given two firewall servers, A (primary) and B (failover) both running ntpd
service available to clients on the LAN. pfSense 2.2.1/amd64 on pfSense
branded C2758.

The following procedure more or less causes ntpd on B to become unstable:

1) on A, set CARP maintenance mode to make B answer all CARP interfaces.
2) reboot A.
3) clear CARP maintenance mode on A to return all shared IPs to A.

Monitor ntp on both firewalls. After a while, A is still pretty stable and
has normal status. On B, however, the "reach" value for the upstream ntp
servers drops down to near 1, sometimes 0, and the offset gets into the
hundreds of ms, very close to 1 second and the jitter is very high. It does
eventually recover and then flip back and forth between accurate and barely
accurate time keeping. It also appears that the process dies frequently,
but the watchguard package restarts it for me. Perhaps that restart does
temporarily fix the time keeping.

The only solution I've found to get it back to keeping good time is to
reboot the server. Simply restarting ntpd is insufficient.

Can anyone else reproduce this or has seen it?
_______________________________________________
pfSense mailing list
https://lists.pfsense.org/mailman/listinfo/list
Support the project with Gold! https://pfsense.org/gold

Reply via email to