I think it's just the CPU crystal, these are not Stratum clocks, or maybe they are Stratum 4 or something. If you are looking for something to maintain accurate timing if ALL the masters become unreachable, this is not the solution for you, then you need something like a couple posts back in the thread.
I'm just saying don't configure every NTP client in your network with 4 external master NTP servers out on the Internet. Use routers or Linux servers on your network for that, and sync all your clients to them. In my case, if none of my core routers have connectivity to the Internet, I've got bigger problems than NTP. And as long as my whole network is using the same on-network NTP server, it probably doesn't matter if it's a few seconds off. -----Original Message----- From: Af [mailto:af-boun...@afmug.com] On Behalf Of Chuck McCown Sent: Monday, October 10, 2016 10:43 AM To: af@afmug.com Subject: Re: [AFMUG] Update on DHCP - Calix failures What do they use for a timebase? -----Original Message----- From: Ken Hohhof Sent: Monday, October 10, 2016 9:40 AM To: af@afmug.com Subject: Re: [AFMUG] Update on DHCP - Calix failures If you have Cisco core routers in your network, I have found them to be good NTP servers. Set them up with however many masters you want, there are nice ACL commands to limit clients and prevent amplification attacks. Should probably specify them by loopback address to insure they are reachable even if OSPF reconfigures path to router. -----Original Message----- From: Af [mailto:af-boun...@afmug.com] On Behalf Of Chuck McCown Sent: Monday, October 10, 2016 10:21 AM To: af@afmug.com Subject: Re: [AFMUG] Update on DHCP - Calix failures Nice clock source. Thanks. I think I am going to buy the larger one. It has an onboard OCXO for holdover timing. -----Original Message----- From: Seth Mattinen Sent: Monday, October 10, 2016 8:55 AM To: af@afmug.com Subject: Re: [AFMUG] Update on DHCP - Calix failures On 10/10/16 7:32 AM, Chuck McCown wrote: > I am guessing that Xmission was giving us a malformed or otherwise > unpalatable NTP time hack. Calix reacted by setting the RTC to goofy > time. I have never dug into the internals of NTP. I guess the next > step would be to wireshark that and figure out what is going on. But > Calix should fall back to secondary NTP or freerun if it does not like > what it is getting. And give an alarm. How many tickers were configured? Good practice is to configure at least 4 from different sources so that NTP can detect a false ticker or bad time source. Configuring two is bad. If you only have the option to configure one or two NTP servers then only do one but sourcing off a time server that has at least four. NTP doesn't do "primary" or "secondary". These are cheap to have your own GPS clock: https://www.css-timemachines.com ~Seth