n1 is a stratum 1 server steered by pair of GPS receivers.
n1 exists because we need a reliable source of time. One where I know it'll either give correct time,
or be honest and say it is unsynchronized. I've been burned one time too many.
2015.3.21: Failure of GPS #1 and #3
Both RF Solutions GPS-1513R receivers failed at exactly 0800 UTC. Now running on the #2 checking receiver.
2015.2.15: Workaround routing to the incompetent-sysadmin-bogon-list is down. That was through vp in .nl
For those running faulty and out of date bogon lists, your own fault. Fix it on your end.
2015.2.15: raid failure on ih nl node 3. Which is bad; we are down to one working outbound IPv4 MX
Yay. More work.
n1 (dk IPv4/6) working
fr (dk IPv4) working
rosetta (dk IPv6) working
vp (nl IPv4/6) FAILED
vix (nl IPv6) FAILED
us (usa IPv6) working
mx15 (dk IPv4) down for service
2014.10.18: Loss of GPS2 for 8 hours.
gpsd lost GPS2 at approx 0400 UTC. PPS was still coming. Nothing in the logs.
2014.08.05: Loss of GPS2 for 11 minutes
gpsd lost GPS2 at 7:23. PPS was still coming. This is either the BMC messing with the port, or a gpsd bug.
2014.04.15: Web server fixes
During the server move, I did a nuke and pave on file permissions and ownership. That meant that some directories were unreadable to the webserver.
2014.04.15: Upped the rate limits.
For some experiments. And to see what the abusers do.
The limits are still low enough to prevent reflection attacks.
2014.04.14: Back in spec.
Kernel PPS is now fixed. Offset is verified, jitter is low, good. A traffic plot and skyview plot was also added. Note: This hardware has 15us pulse to PPStime
2014.04.13: No more IE 8 on XP, no Java 6.
The server now enforces forward secrecy on encrypted connections.
2014.04.13: Normal Operation.
The replacement server is fully configured, patched for heartbleed, duct taped, and has seen the first 600k clients.