Björn Gabrielsson wrote: > On Fri, September 14, 2007 4:15, Ask Bjørn Hansen said: >> Being 100ms off for short intervals is okay for clients using sntp >> (presumably they don't want much accuracy) and ntpd will probably >> pick another server to synchronize with, but your server is still >> helpful in deciding the correct time if one of the other configured >> clocks is going completely nuts. >> >> >> - ask > > I doubt _very_ much that Tims server is 100ms off for clients in > Sweden/Europe. His server is syncing to Lysators pool-server (and other > good local S-1 servers at offsets <<1ms. The penalty comes from having a > bad route to the pool monitoring system -- not from serving bad time. >
OK. Now I have done some traceroutes from my server to x3.develooper.com and have found that a congested link is the cause of the problem. This morning, when the monitoring system showed a -130 ms offset, the UDP roundtrip from my server to x3.develooper.com took about 500 ms. Now, when the monitoring system shows a -5 ms offset, the UDP roundtrip takes about 190 ms. The problem seems to be between Denmark and the US. This morning hop 10-12 from my server looked like this: 10 pos3-0.2488M.nyk2nxg1.ip.tele.dk (83.88.12.141) 117.571 ms 117.587 ms 117.454 ms 11 * * * 12 cr2-cr1.nym009.internap.net (66.79.146.198) 398.856 ms 403.317 ms 403.627 ms and now it looks like this: 10 pos3-0.2488M.nyk2nxg1.ip.tele.dk (83.88.12.141) 117.751 ms 117.637 ms 117.563 ms 11 * * * 12 cr2-cr1.nym009.internap.net (66.79.146.198) 117.825 ms 118.071 ms 118.013 ms /Tim
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
