n1 is a stratum 1 server steered by pair of GPS receivers.
n1 exists because I 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.
2022.10.31 A Quarter of the world - yay?
This server has now seen a quarter of the IPv4 address space as clients. The first 600 million addresses went quickly enough, then it took somewhat more time. IPv6 is now 25%, total is about 240 terabytes of NTP traffic. Shoutout to Vatican City for using me as primary time source
2022.05.22: Spoofing, and SPF+DKIM+DMARC
Someone is spoofing the taur domain, and presenting this host's name in the handshake. So the SPF rejects and DKIM failure are to be taken seriously.
Hooray for dmarc reporting.
2020.03.25: Fibre is good for the tummy
The cause of the receiver resets is a combination of a firmware bug and the characteristics of a local noise source.
The plan is to switch to a different receiver - until then, the secondary receiver is going to be the master.
The secondary receiver is now also an MC-1010, but it is located in the roof of building 10.
A bit of encoding magic drives PPS and TXD over the same fibre with very low jitter. From the ceiling space, the first
stop is server room 10, it gets patched through and ends up at the building 10 exit point. It gets patched out, and
arrives in the building 8 switching room, gets patched again, and ends up in server room 8, with who knows how much loss
due to the ebay-quality patch cords I used. There it gets decoded and drives N1.
All is not well though - I know that I have 16us of transmission delay, and it's not subtracted out yet - ntpd is being cross with me.
So, sadly, I'm 16us off. Not that anyone outside is going to care, but I do.
2020.10.30: Random guess
The 3.3V regulator in front of the GPS receiver is well behaved, low noise, verified good, and highly reliable type. But it is also the only part of channel 1
I have not replaced. So I have now replaced the known good regulator with a hacky zener-transitor regulator. And while at it, removed the
2020.10.22: New serial card
The server is suffering resets in the primary GPS receiver. I have replaced the receiver, the line driver, the cable, and I still have the resets.
So, in went a new serial card, timer-capture module, and power supply for the line drivers on the server end. They are the only things I have not yet replaced.
Funny thing, those two cards cost more than the rest of the server. No resets yet, but I will know in a month or two.