but there are problem with it - reach value is 102, which means that it is unstable. All last values are: 241, 120, 250, 124, 52, 25 - and all of them are unstable, which means that this server is a bit of problem, but it still can be used as reference. I suppose that we should in long term improve our algorithm to detect problematic servers (for example, pass unstable references if there are more than one second references) but in short term we can just remove requirement to discard unstable one, as it still used for time sync.
If we look into last snapshot logs, we will see that there was a server reached:
> *81.2.248.189 195.113.144.238 2 u 75 64 102 7.526 -0.880 0.874\n
but there are problem with it - reach value is 102, which means that it is unstable. All last values are: 241, 120, 250, 124, 52, 25 - and all of them are unstable, which means that this server is a bit of problem, but it still can be used as reference. I suppose that we should in long term improve our algorithm to detect problematic servers (for example, pass unstable references if there are more than one second references) but in short term we can just remove requirement to discard unstable one, as it still used for time sync.