I have the same issue from Belgium. IPv4 address stays in negative and dropping. IPv6 address has perfect score. Tested everything with 2 different sites, one of which you also used, and both report my server is doing fine
First post here…
I had my server ntp1.mgrey.se +2 years without any issues always keeping a score above 10.
Last 2-3 weeks started to get a lot of ‘i/o timeout’ from Los Angeles station (every hour).
I can not see any issues with my ntp server from outside using web tools and other ntp clients.
ntpdc -c sysstats localhost gives:
Only about 25-35 packets/s, but I also see about 13,5% ‘rate exceeded’
(i have discard average 5 minimum 1)
We have the same problems in Finland at MIKES.
The attached image shows number of IPv4 users for our public stratum-2 servers, which has fluctuated wildly since mid-February 2019 because the Los Angeles monitoring station produces erratic scores for us.
If there’s a mechanism for setting up a monitoring server we could contribute - but I would need detailed instructions.
Even adding it to the pool was unreliable; it first showed “Could not check NTP status” but worked the second time. I just got the same error trying to add it to the beta (is that still a thing?), which seems to be from LA as well. I’m inclined to just remove this server and not worry about it, but I can’t quite account for why I’m not seeing problems in my own monitoring, or why AWS’s network would be so unreliable without wider objection.
Traceroute was not too informative; AWS drops ICMP at most of its routers, but in general the path looks to be AWS - Seabone - NTT - Phyber outbound, and possibly straight Phyber - NTT - AWS inbound.
Thanks! I ended up scheduling it for deletion, because it’s consistently had a score below 10. I double-checked that conntrack wasn’t secretly running and torpedoing connections, either. The monitoring server just hates this server.
I have a server in Chinahttps://www.ntppool.org/scores/18.104.22.168. It gets the same problem. The score never passes 10 after I joined it in the pool. Actually, most of the time, the score is below 0. But I don’t think it has any problem from my own monitoring result. Does anyone know why this happens? Thanks.