As we have server pages, may we have public monitor pages?
A kind of reverse data relative to the server page: for a given monitor show the list of monitored servers. Probably with the following detailes: Active, Testing or Candidate mode, and the score and RTT of each server from the given monitor.
At the very least it would be interesting to see which servers my own monitor is monitoring, without using tcpdump. I’d find it motivating to see what it does, beyond just an aggregate of “140.7 tests/min (135.2 ok, 3.8 timeout, 1.7 offset)”.
Not necessarily public, but in the monitor status details.
Personally I do not care about this.
But it would be funny to know how a monitor scores others.
And based on that get’s it’s own public score.
I mean, a bad monitor below 20 should be kicked out after a week.
And a good monitor scores 100.
Meaning you can trust it.
Same as scoring for NTP-severs, score Monitors.
And monitors that keep scoring bad after 1 month get an email, telling them to consider to stop as monitor.
So they no longer hog the list.
When your monitor keeps scoring below 20…it should be taken out of the active-monitors, stay bad…consider removing it in total.
Also publish what zones THAT monitor was testing…there could be a pattern.
My 2 cents.
Hmm, my understanding so far had been that every monitor is monitoring every server out there (with exceptions due to how monitors are phased into the system and need to work their way “up” to show up in the table/graph, see below). The main difference is the frequency at which they poll the servers, based on the “category” a monitor is in for a specific server. Nicely visible in the graph of a recently added server, before the amount of data makes the dots clump together.
The difference in polling frequency is also why new monitors take a bit of time to appear in the table/graph of existing servers in my understanding. Because they start with the lowest polling interval so it takes longer for them to generate data needed for them to move “up”, i.e., have the scores for servers increase from the initial value.
I think it’s a bit different when a new server is added within a small time window around a new monitor having been added to, and accepted into the system. Because then, the primary criterion for assigning initial “Testing” monitors to servers is not how well they score (none of them produces a good score yet, not even the more mature monitors), or has a history of scores for a newly added server), but rather the RTT to the server. I.e., a newly added/accepted monitor has a similar chance of getting into the pole position with a newly added server as a more mature monitor, while for an older server, all (or most) of the spots in the higher categories of monitors have already been taken, and the system on purpose is slow to swap out/in Active and Testing monitors.
EDIT: On a newly added IPv6 server, I am seeing 92 monitors in the table, and 96 hitting the server. While I don’t know for sure whether the difference is due to some monitors added to the system but not yet accepted, that seems like a somewhat plausible explanation. And would support that pretty much all monitors monitor all servers, though one would need to check on more servers to establish whether that is indeed the case.
Anyhow, if that is really the case, listing all 3600+ IPv4 servers and 2100+ IPv6 servers for each monitor would need some innovative way to present that amount of data on a web page. But getting the info as a machine-readable file for offline evaluation and analysis could be easier to realize, at least in a first step, e.g., for assessing patters regarding where the monitors reside that monitor a server in a specific location.
You do know that assigned monitors hit harder then the rest?
As I checked the monitor page of mine, and assigned monitors makes a lot more requests a second then others.
Before they where assinged, they were only doing 10 or so tests a minute.
At FULL power (assigned) they test ALL servers several times a day.
BTW, I takes weeks to get acepted as an active-monitor.
You are not accepted instantly based on your friendly face 
Yes, that is why I wrote the following:
Sometimes, it is real quick, like just a few days.
No, as it is a manual process that apparently only Ask can do, it depends on when, and how often, he gets around to it.
Exactly…and that became a problem for me…and tossed so many routers.
As tables ran out.
The MikroTik can handle it, so for the moment I run IPv4 monitors only.
May change to dual-stack…not sure yet.
I thought from discussions in other threads that by now, you had understood that it were not the monitors’ packets that were causing your issues on the router, but the clients’ packets… 
That is the thing. I still do not know.
Even the DrayTek (60K sessions) slowed down.
Routers and NAT do not like UDP traffic in high quantities like we do.
Yes I still get high requests, but the MikroTik isn’t bothered…it seems.
And the monitors seem unaffected too.
So I can (at this time) conclude that many routers are unable to handle massive NAT-tables.
We will see…I have it running for about a week now, it seems fine.
IPv4 and IPv6…fingers crossed.
NTP traffic isn’t a huge load, but fills the tables…and I think now, the table-limit is the problem.
You still don’t know that it is extremely unlikely (though almost never completely impossible) that it was the packets of the local monitor you operate that cause(d) your issues?
Ok, off-topic in this thread, so won’t pursue/reopen that discussion here, or at all… Except for wishing you good luck with the new router! 
You have to know that e.g. a DrayTek has 1024-entry NAT-table limit.
That is filled in seconds with NTP as it’s UDP.
For a simple homeserver this is not a problem.
But we get 100 req’s a second? So the table fills fast and has no time to empty.
I did enable the DDOS detection on it for 123, it was triggered all the time.
Simple routers simply can not handle this type of NAT-table-flows.
It’s running for a week now, I have not seen slowdowns.
BTW other routers I set the speed to 512K, I have set it for testing to 6M, I only saw the traffic go up, but it didn’t slowdown my own internet-usage.
I have never ever set it that high before.
You know my complains before…I’m starting to think it all comes down to poor routers, where they are promoted as FAST consumer routers. But they are not when it comes to our usage.
So far the MikroTik holds the castle 
All those data is already available for public. It is just organized differently to present to the users.
On the server score page the name of the monitor could be a URL to lead to the monitor’s public page. An IP or the name of the server on a monitor public page can be a URL leading to score page of the IP or name.
It would be nice to be able to click on a monitor.
Then see where it’s good and where it’s a mess.
Show countries where it’s wrong.
Or simply show how many servers it’s wrong/not able to reach.
At the same time, score the monitor where it does reach and work well.
I do believe it will help to check/show us if MY server is to be trusted or crap.
I think it will help to show bad monitors to us all, as well as good monitors.
BTW a bad monitor doesn’t mean you are a bad person! Just you have poor Internet.
Shit happens.
