Julian Fietkau @julian@fietkau.social
Public

Okay, so either Deno's OpenTelemetry metric “http.server.active_requests” (docs.deno.com/runtime/fundamen) is actually cumulative over the lifetime of the process, or I have found the explanation for why I can't get the Encyclia server to behave.

Next question is what I could be doing wrong. 🤔 All it does is “deno run” and everything that doesn't time out seems to return what it should.

(The middle of the graph is where I restarted Deno twice.)

Line graph showing "http_server_active_requests" over a 9 hour duration. It starts at 0, climbs to 5000 almost in a slow straight line over 3 hours, then drops back down to 0, makes a tiny, quick bump up to 100 and back down to 0, only to start climbing again, slightly slower, to 3000 over the course of about 3.5 hours
Julian Fietkau @julian@fietkau.social
Public

Ah, found the culprit. It was one of the spots where I told the program to do nothing later. Turns out my wording was slightly off.

Before:

return new Promise(() => {});

After:

return Promise.resolve(null);

The number of concurrent connections looks negligible now, and no more timeouts either. Calling it a success for tonight, we'll see how things look in the morning. 🥱

As underwhelming as the cause is, that journey might deserve a blog post if I find the time!

0
0
0

If you have a fediverse account, you can quote this note from your own instance. Search https://fietkau.social/users/julian/statuses/115403919822882930 on your instance and quote it. (Note that quoting is not supported in Mastodon.)