Hello, On 31-10-16 18:07, Miroslav Lichvar wrote: > On Mon, Oct 24, 2016 at 02:29:25PM +0200, Attila Kinali wrote: >> I wondered why there is only one monitoring station in LA. >> Is there a special reason that there aren't more?
What would you want to accomplish with multiple monitors? > I was wondering that too. I guess one problem with multiple monitoring > stations would be difficulty in combining their results. That is the main thing. Monitoring is not about detecting if a specific server locally keeps the right time, although this seems to be forgotten very often. Monitoring is about detecting if a specific server serves "good enough" time to some client somewhere on the internet. So, what would you do if one monitor does not receive valid time from a specific server while another monitor does? Is the server able to serve good time to any client? The most reliable thing to do is to raise the score of a server *only* if it is reachable by *all* monitors (and else decrease it). In that case multiple monitors enable to detect problems in multiple routes. With the current one monitoring solution only one route is checked. However, is this worth all the trouble? I also guess this is not what many have in mind when they ask about multiple monitors. It would more often result in a lower score than with a single monitor. > Will they be > managed by trusted people or random volunteers? How many are needed? > I suspect simple voting would not work well with servers that are > reachable only in their country zones (e.g. China). As far as I know, all software used in the pool.ntp.org system is publicly available. As the pool is only about resolving DNS queries, it is "easy"(*) to set up a totally separate pool using a different domain, like pool.ntp.cn. That is, if somebody wants to put the effort in it. If an NTP server operator likes, he can join many separate pool projects. I think this is the best option for isolated area's. However, somebody from within that area has to take the initiative to set up the infrastructure and install the software. (*) "easy" as nothing special has to be done to the software of the current pool system (as far as I know). > I was thinking maybe the NTP servers themselves could be monitoring a > random selection of servers from the local and neighbouring country > zones. A script could be used to periodically upload the reachability, > offset and other data from the NTP daemon to some location created for > each NTP pool account. The question is how should be reports from > different servers combined. What is the problem with the current pool, or what problem do you fear to experience in the future? Kind regards, Arnold _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
