In Israel today:
I doubt this is a pool problem.
As there should be many peers going in and out your contry.
If some major distribution system is down in Israel, it could be the problem.
You may want to check your ISP what’s going on.
Edit: I have checked this site…
Where all peers come together at IIX…seems to me a lot of lines do not carry any traffic at all.
This was a problem with the pool, there was only one server in the Israel zone so if you used pool.ntp.org from Israel it would return that one server. The problem is that that one server was offline( it was added by an unauthorized individual per the irc channel) but the monitor was not updating so it stayed in the pool.
FWIW that server’s score is still showing as over 10, so it appears the monitoring problem persists.
I wonder if it wouldn’t be wise to promote country requests (or requests from a given country) to the containing continent zone when there are less than 4 servers in that country zone, rather than only if there are zero servers in the country zone.
I understood that @ask is planning to move away from country or even continent zones anyway, see this thread. So we need to see whether he opts for tweaking the current country zone-based mechanisms before making those zones just aliases to the geolocation-based automatic matching system, or whether he’ll focus on the new approach right away.
Though some quick fix to avoid the starkest negative effects in the meantime, like with the Israel zone in this case, certainly wouldn’t hurt.
Yeah, that is kind of annoying, is independent of the zoning scheme, and hopefully gets fixed before the new approach is in place, see another example below:
All except one monitor scoring well above 10, majority of active ones even well above 19, but overall score at -99.9.(Though one could argue in this case that the server having been unreachable for several days in the recent past and another hiccup since then might make the scoring algorithm more cautious in restoring trust, but I don’t think that’s it, seeing all the other examples.)
Ask told us the database-server ran into query problems where it never finished the query.
Should have been solved by now.