Does anyone know if there are many pool monitors in AWS

Server operators
1

Hi

Supplementary question - Would it help, as in change what I am seeing, if there were more in AWS?

I am seeing three servers in the pool (there may be more) that are always showing up in the servers giving the most connection errors.

https://www.ntppool.org/scores/2001:df4:bac0::e35b:f223 plus https://www.ntppool.org/scores/2a06:a880:3:bb1e::123 plus https://www.ntppool.org/scores/2a06:a880:6:2090::1234:

Looking at my results deeper I am seeing that a non AWS server in a location normally has few if any connection errors while a server in AWS at the same location has quite a number of errors when connecting. This could of course be an AWS issue from multiple regions to multiple pool servers - Internet routing is a strange place - but it “seems” to be mainly problems connecting from AWS ip addresses to some pool servers.

$ /usr/local/bin/dump-results-2-stdout.sh 3 | grep -v Tested_CAC | grep 2a06:a880:3:bb1e::123 | awk -F, '{print $2,$3}' | sort | uniq -c | sort -k 3 | egrep "sin|hkg|bom"
     14  Tested_OK  cn-hkg-aws
     35  Tested_ERR  cn-hkg-aws
     40  Tested_OK  cn-hkg-lightnode
      2  Tested_ERR  cn-hkg-lightnode

      9  Tested_OK  in-bom-aws
     43  Tested_ERR  in-bom-aws
     44  Tested_OK  in-bom-vultr
      1  Tested_ERR  in-bom-vultr

     10  Tested_OK  sg-sin-aws
     34  Tested_ERR  sg-sin-aws
     57  Tested_OK  sg-sin-ovhcloud
      3  Tested_ERR  sg-sin-ovhcloud

Over the last 3 days for example Mumbai AWS has tried (9+43) 52 times with 43 errors while a server in Vultr in Mumbai has tried (44+1) 45 times with only 1 error.

I see that on one of the servers there is (at the time I am looking) one monitor with -100, one below a socre of 10 and all the rest are reporting above 10 to give a current score of 20.

The supplementary question is if there was one or a few monitors in AWS, giving poor results then would the servers score actually be reduced that much or would the majority reporting good results out vote the minority.

Now I know that the current situation is way better than having just one monitor and I also know that the monitoring system can’t test from every ip to every pool server but I am wonding if there is a way to improve things so servers that seem to be having problems receiving requests from a particular location or hosting company get highlighed.