[TriLUG] Speaking of ntp: that would be a great service for T rilug to provide.

[Jon Carnes]

I agree 100% with Jaimie, and I've had the same kind of problems (old
servers that I used to use now being restricting their access).

TriLUG servers already sync their time as part of the NTP network. All
I'm asking is that we modify one of the servers to allow folks to query
that server for time.  That's like a 10 minute configuration change. 

<not a lot of work for a big return>

If you want to heap warnings on the use of that service, that is fine
with me. Folks *should* use more than one source, especially if time is
important to them.  For my home machines, I'll just point them to
clock.trilug.org and be happy as long as I can get a good reading once a
day.

Jon

On Tue, 2003-04-29 at 12:48, Jaimie Livingston wrote:
> I am only aware of 1 public Stratum 1 (terrapin.csc.ncsu.edu) and 1
> semi-public Stratum 2 (clock1.unc.edu) in North Carolina. I currently use
> the Stratum 2 servers at VT-Blacksburg in VA (ntp-1.vt.edu and
> ntp-2.vt.edu). I've found a few other NTP servers in the US that I can use,
> but I've found that most Stratum 2 server operators are now implementing
> regional or community access restrictions. If there are other open public
> servers, I didn't find them in the Mills' Public NTP Lists, but I also
> havn't look closely in a few months.
> 
> Wouldn't it be within TriLUG's charter to provide a Stratum 2 server to the
> community at large, not just TriLUG'ers. It would be a reasonable service to
> provide, and has the potential to give the group a larger exposure to the
> business community that would take advantage of the server. Assuming that we
> have somewhere to host the server...
> 
> Just a penny's worth...
> 
> Jaimie
> 
> 
> -----Original Message-----
> From: Mike Johnson [mailto:mike at enoch.org] 
> Sent: Tuesday, April 29, 2003 12:14 PM
> To: trilug at trilug.org
> Subject: Re: [TriLUG] Speaking of ntp: that would be a great service for
> Trilug to provide.
> 
> 
> Jon Carnes [jonc at nc.rr.com] wrote:
> > Don't have a cow man!  To me trilug is a trusted source.  I trust the 
> > folks running the services to do a good job and in setting up NTP so 
> > that it works great.
> 
> Trust isn't really the issue, here.  We provided mirrors because it's
> difficult to find high speed mirrors that aren't full.  There's a ton of low
> latency public NTP servers out there, and it doesn't take long to track them
> down.  It would be redundant (and inaccurate) for us to set up a stratum 3
> server for people to sync off of.
>  
> > As an example, I had no idea there were tier 1 or 2 or even 3 ntp 
> > servers.  I think it's cool, but really, do you think every TriLUG 
> > member wants to do a few hours of research in setting up their ntp 
> > service - or do you think we might want to just point to a trusted 
> > source and be done with it?
> 
> No, but if every TriLUG member does a few quick google searches, they'll
> find the NTP servers.  And they'll read about it, and -learn-.  Time is
> -not- simple.  Time is important.  People need to learn, and understand
> about time.  But you do it once.  You sit down, learn, and then you know
> about the different levels of time.  And you know where to go to get a
> server list.  And you come up with three or for low latency servers that are
> open access, and you write them down in your notes, and use them.
>  
> > Also, wouldn't it be of more benefit to the NTP network structure if 
> > the load from all us individual users were off loaded to a local 
> > server?
> 
> Possibly, but trilug.org isn't neccesarily 'local' to everyone within
> TriLUG.  I would imagine that if you ping some of those servers in the
> stratum 2 list, you'll get lower latency than to trilug.org
>  
> > I'm sure that Trilug uses more than one Tier 2 source in syncing their 
> > clocks, and so is fairly accurate, probably to within a tenth of a 
> > second.  Well that is way more accuracy than I need.
> > 
> > What I need most of all is easy setup of my NTP service - a single 
> > server that I can simply depend upon to give me reasonably accurate 
> > local time.
> 
> But you cannot rely on one source for time.  If that one source goes down
> (as happens every now and again), you loose sync.  If you loose sync and
> drift too far, you'll end up with large jumps in time, which can be bad
> depending on your services.
> 
> In the end, if there's enough interest in it, I'm sure we'll see what we can
> do.  But I do not think it's a good idea.
> 
> Mike