Monitoring metrics

How are the totals in the light blue calculated? I have only the two monitors and if I add their reported 1 hour tests/min, I get 13.1 + 9.6 = 22.7, which is roughly half of what is reported in the blue block. The 24 hour rate in blue is also roughly double. Is it because the monitors have both IPv4 and IPv6 addresses?

I moved this from the “mega thread” so it wouldn’t get lost.

The metrics are from prometheus, and yes – something is definitely off!

2 posts were merged into an existing topic: “monitor=recentmedian” no longer works

This has been fixed!

1 Like

No, it doesn’t seem to have been fixed. It’s still showing exactly double.

“My Stats” are correct. The sum of all monitors is equal with the one from the blue banner.

You are both right, I think! Some of the accounts are still showing the double counts. :thinking:

Both the production and beta sites still show double the sum in blue for me. All my monitors are registered for both IPv4 and IPv6.

I see the values for ok, timeout and offset disappeared on both the beta and prod systems?

image

:exploding_head: Right this moment it looks okay in your account; your two monitors numbers add up to the total in blue.

The ok/timeout/offset box is supposed to show how many results of the various kinds, but only if they are more than 0 (and it’s not even showing 0, so that’s not the bug). That feature looks completely broken, I’ll have a look later. Maybe I lost track of what bugs I was fixing and pushed an update to the website but not the API behind Sunday evening. I haven’t touched the production site all week.

Is it possible that it can show different for different people? Even after a reload of the page, I still see double:

image

zakim1-yfhw4a and zawkf1-yfhw4a

imageimage

I see the totals are not double anymore and the ok, timeout and offset fields have numbers again. Thank you!

image

At least my dashboard currently is still missing the ok/timeout/offset breakdown numbers.

It was fixed on the beta site for me. I think it has not been rolled out to the production side yet.

Ah, ok, sorry. Since the rollout of the new monitoring to the production site, I admittedly haven’t paid much attention to beta anymore, also given issues in adding new monitors there (which Ask is possibly going to look into later today). I.e., I don’t know what the situation was in the interim, but now looking good wrt ok/timeout/offset on beta:

Screenshot from 2025-08-10 19-53-21

1 Like

Rates doubled, no breakdown numbers for me.

I’ve updated the production site.

4 Likes

The quorum mechanism now seems to no longer allow a single monitor above the threshold to supersede all other monitors below the threshold. At least in the situation shown below:

1 Like

I am sorry that conclusion seems to have been premature:

1 Like