Minor questions from observations of new monitoring system

I understand the new monitoring system is still evolving. Seeing some behavior, I am just wondering if that behavior is as intended. I endeavored to follow the discussions, but apologize in case I missed that an aspect has already been answered.


With respect to the two figures:

  • I understood that while not used for scoring, monitors in the “testing” category would nonetheless be pruned in case of suboptimal interactions with a monitored server. Still, while it apparently has no connectivity at all to the server in the figure (e.g., it is known that there is no IPv6 connectivity between HE and Cogent), monitor usday2-1tcp71g is being kept around as testing monitor. (It was swapped out for about two days just before the time slot covered by the figure, but had been in the list for several days before that already.)
  • Related, while I understand the active monitors should give a good “overall” picture of a server, i.e., not only monitors giving “good” scores should be included, in this case, I would have thought that one or more monitors with higher scores than shown in the figure could have been included in the active monitors instead of being in the testing category.
  • Why isn’t the overall score shown in the table? EDIT: Just notice that more servers are not showing the “overall score” line in the table anymore, so I assume I have missed mention of it being dropped going forward. (Overall score is still given above the offset/score graph.)
  • It seems two monitors can generate a minimal value of -80 (usday2-1tcp71g, see figure, but also at least usewr2-1a6a7hp, seen on another server of mine), while the majority has a minimal value of -100. Is that intended, e.g., temporarily different software versions?


1 Like

I am seeing the same problem and don’t know really what is happening:

Here is another one…
Score is 0.0 even all of the active monitors are > 10.0

And a “reversed” score :slight_smile: All the monitors show a negative score but the overallscore is positive

@ask can you take a look please ? :slight_smile:

1 Like

Here is yet another one…
The score is 0.0 even though most of the active monitors are > 10.0

1 Like

1 Like


I believe the monitors 23 (every) and 24 (recentmedian) broke recently. I used them for monitoring offset and score.

Last “every” result I received is from 16th of August:
1692222258,“2023-08-16 21:44:18”,-0.002420517,1,19.9999970578419,23,every,0,
Last recentmedian is from 25th of August:
1692971255,“2023-08-25 13:47:35”,1,19.5046767891312,24,recentmedian,

This would explain the discrepancy between individual monitors and the total score.


1 Like

If you look at it, all reported problems are Israel and not other countries.

Looks to me IIX has problems with peers or routers.

Sorry, that is not true. My servers are in Germany, and the last post by @jord903 seems to concern a server from Taiwan.

Also, at least the issue I raised is not that monitors are giving non-optimal scores for servers. But how scores from monitors are aggregated to overall score, and how active monitors are being selected. I would consider it unlikely that the country a server is associated with has anything to do with how the monitoring aggregates scores or selects active monitors.


I don’t see how your post relates to the Israel issues, they drop to below zero.
Yours doesn’t.

I mean, not as far as the monitors go…the calculated number is wrong or wrongly presented.
Report this to @ask in a different topic. Thanks.

Sorry, but I think you are in the wrong thread. This one is NOT about the Israel issues just because some servers in context of this thread are from Israel.

I think the thread about the issues with the pool in Israel is this one.


Here is another on from the US:

1 Like

The same problem - I set up two NTP servers - the first one was added and after 24h I got a score of 20. The other server was set up with the same IP prefix on a similar hardware server, and my score is unwavering at 0.0.

I will be happy if anyone can diagnose this issue.


1 Like