I’m seeing incorrect app concurrency metrics with symptoms pretty much exactly matching those in this thread I posted last month.
Apparently it’s been drifting from the true concurrency count sporadically for (at least) the past two weeks:
Restarting the affected machines temporarily resolves the issue. Here is the fly.toml.
@bglw you looked at this issue last time it occurred so I’m tagging you in case you are interested - sorry for the ping otherwise 
I am interested indeed, since this should have been resolved by the last fix! Sorry about this.
We’ll re-investigate what’s going on here and get back to you.
Hi @tjhorner, just to keep you in the loop: I think this is likely a regression of a bug we had before, where each external port exposed by your app isn’t distinguished in the metrics, causing this oscillating behavior (they’re essentially 2 metrics being scraped as 1). We presumably fixed this in a Fresh Produce long ago, but a recent change likely caused this to regress. We’ll push a quick update to fix this soon.
Thank you, appreciate the update and additional details!
Hiya, any update here? The issue is still present on my end.
Hmm, I thought I had fixed it last week by reverting the regression, but it does look like the metric on one of your machines is oscillating between two values again. I’m taking another look.
@tjhorner After a bit of rubber-ducking today I think I have tracked down the root cause of this issue and the fix has just been deployed. Let me know if this shows up again!
Thanks! Will monitor and shout if anything looks awry. Looking good so far.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.
