davy@itstd.sri.com (02/27/90)
Does anyone have the appropriate "driver" to make either ntpd or xntpd talk to a Kinemetrics TrueTime Model 60-DC WWVB clock? We sort of found one lying around in a cabinet here, and want to plug it in and make it available to the world as a stratum 1 clock... Dave Curry SRI International
Mills@udel.edu (02/27/90)
Dave, Right on! With another GOES clock, maybe we can measure the yank of the San Andreas fault. I know of no such driver, but maybe how would you like to tackle that feat? Feet? Dave
davy@itstd.sri.com (02/28/90)
From: Mills@udel.edu
Date: Tue, 27 Feb 90 20:56:34 GMT
Subject: Re: clock driver?
Dave,
Right on! With another GOES clock, maybe we can measure the yank of the
San Andreas fault. I know of no such driver, but maybe how would you
like to tackle that feat? Feet?
Dave
Well, this isn't a GOES clock, it's a WWVB clock (Model 60-DC, not 468-DC).
Apparently we purchased two of them for some project a few years ago,
and this one's been sitting in a cabinet ever since the project ended.
Nobody knows where the other one is. I even had to call up Kinemetrics
to get a manual for it. They were very nice and mailed me a free one,
though. The guy I talked to had never heard of NTP, but I told him a
little about what it did, and he sounded interested. Unfortunately,
he didn't include his business card with the manual, so I can't mail
him any more info. Oh well.
Anyway, if nobody has a driver, I reckon I'll be writing one. I
haven't decided which program (ntpd or xntpd) to write it for yet,
though. Before I found out we had a clock I was planning to use ntpd
everywhere since xntpd seems to be sort of large to run on diskless
machines. But now that I have a clock, I guess I'll have to rethink
all this. Sigh.
If anyone has any recommendations, suggestions, things to watch out
for, I'd appreciate hearing from you.
--Davebob@kahala.soest.hawaii.edu (Bob Cunningham) (02/28/90)
Just sent you the source for read_tt468.c (for ntpd). Check the format of the input it expects (which is that of a GOES clock) and if it's like what the 60-DC uses you might be able to use it. If not, then perhaps there's enough similarity that you might be able to use it as a template for your own code. You'll want to check both the exact format assumed to be provided by the clock as well as how "quality" is handled (which I'd guess might be considerably different). To actually use it, you'll need to fiddle with ntpd.c and Makefile a bit. I'll be glad to supply the simple diffs I use for those to get read_tt468 working if you like.
davy@itstd.sri.com (02/28/90)
From: bob@kahala.soest.hawaii.edu (Bob Cunningham)
Date: Tue, 27 Feb 90 14:28:18 HST
Subject: Re: clock driver?
Just sent you the source for read_tt468.c (for ntpd). Check the format
of the input it expects (which is that of a GOES clock) and if it's
like what the 60-DC uses you might be able to use it. If not, then
perhaps there's enough similarity that you might be able to use it as a
template for your own code. You'll want to check both the exact format
assumed to be provided by the clock as well as how "quality" is handled
(which I'd guess might be considerably different).
Got it, many thanks. Taking a quick pass through the code, it looks like
making it work with my clock should be fairly simple. The output format
of the 60-DC is the same as the 468-DC, except the 60-DC has a "format"
command which lets me turn on milliseconds, i.e.:
<SOA>DDD HH:MM:SS.sssQ
I don't know whether this is worth having or not; again, if anyone has
experience in these things, I'd appreciate advice.
To actually use it, you'll need to fiddle with ntpd.c and Makefile a bit.
I'll be glad to supply the simple diffs I use for those to get read_tt468
working if you like.
Sure, I'd appreciate them if you've got the time.
Hopefully, at this rate, we'll be "on the air" as a stratum 1 server in only
a week or two (depending on how long I test it for :-).
--Dave