As you might have seen on status.ntppool.org, we’re doing “heavy maintenance” on the web systems this weekend. The website and management portal will be unavailable for a brief period starting shortly.
In progress … These are the old servers, the site is running elsewhere on equipment that is not ok a cart.
Thanks for all your hard work Ask!
Any reason to believe this may have resulted in large weighting changes for some servers? Mine seem to have dropped off significantly in traffic in the last 48 hours (see below). Scores are still 19+.
I added the new monitoring IP around 19.00 UTC (Sunday) and the Los Angeles monitoring system went away a few hours later. It might be back on the new IP since 3-4 hours ago; I honestly forgot to check and I am on an airplane now without enough internet to see. I will check in a few hours when I am home or tomorrow.
Oh, the graphs just loaded and I see what you mean. I will check that the zone data is updating on all the servers when I am back on proper internet. The change relevant for this was indeed about 19 hours before you wrote (and it had been a long week of prep, so maybe I missed something).
The Newark station seems to work like a treat regarding delay and stability. Good work guys!
I can’t get response from https://trace.ntppool.org/traceroute/ by now,
do this maintenance affect the network debugging service?
My offset has changed from -4 (LA) to -0.4 (NJ) in London, UK and score keeps rising (might hit 20).
Early days but looking good, Thanks Ask.
The status page said the maintenance is done, but I cannot add my new servers. I checked it with my other devices the ntp service is working.
The add server page is slow and it said
Could not check NTP status.
Oh shoot. I didn’t do the trace debug tool and forgot that the add server feature uses that, too.
Thanks, Ask, for the Newark monitoring station. Finally after more than two months it seems I have a stable score for my 5 Croatian servers (https://www.ntppool.org/user/zorko).
Actually I was quite worried to get thrown out of the pool, as there are only 3 more servers in Croatia. From Croatia there were no problems, neither it seems to me that other European time requesters (“users”) had any problem getting to my little pool. The score went up regularly enough not to get thrown out of the pool, but each afternoon/night in Europe (i.e. day in Americas) it used to drop and drop…
I’m happy again
PS: I saw that the other 3 servers in Croatia were stable all the time. It may be because my servers are on the Academic Research Network (GEANT/CARnet), and the three others are on the commercial network. This might have been a problem.
Do your graphs show the average rate, or maximum?
There was a change in the DNS configuration. IPv4 servers are now always included in the non-numbered *.pool.ntp.org zones instead of jumping in and out, so the rate should be more stable with smaller but more frequent peaks. The effect depends on how much traffic goes through the non-numbered (SNTP) vs numbered (NTP) subzones. In the zones where I have some servers it seems to work nicely.
If the average rate dropped, it might be due to the switch to the new monitoring node. There are more servers with good scores, so there is less traffic per server.
This has been fixed now; thank you!
It works now, thanks!
My graphing uses LibreNMS, which is based on RRDtool, so I’d assume there’s at least some averaging going on. The differences seem to me more dramatic than the change you describe would warrant. And the server counts in the au and oceania zones (where my systems are) don’t seem to have changed significantly.
If this is the new normal, that’s fine with me, but it seems like there might be some unintended effects happening here. Any thoughts, @ask?
I think a larger number of active servers in other countries, especially if they have high speed setting, could take your traffic at the continental or global level. It would be interesting to see separate graphs for traffic from au/oceania and outside.
My server’s score has gone up to 20 and stayed there since maintenance, a great week. Offset has gone from -4 to -0.1 also. Previous months it was in and out of the pool every few days.
Boab, London, UK
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.