Follow

Finally set up a instance on my vpc, so I can get some nice timing insights into my small running services.

Helps a lot in understanding the whole tracing and spans concept in my projects. With terminal output only, it's hard to figure out the spans as they only show up when you actually log something.

Jaeger was a bit scary because all the docs talk about big multi-instance setups and I have a tiny setup with everything on a single vpc. But the "all-in-one" instance works pretty well.

Just the memory usage is a bit high. Well, it idles around with about 55 MB RAM, which is pretty normal for , but from a perspective it's huge when I see my servers idling around with 1-2 MB, making the tracing collector the most memory hungry instance on my little vpc.

The UI is pretty nice though, so maybe I'll build a tiny rust copy that only implements the tracing variant that I use (as there seem to be several), and whatever the UI needs to function.

Okay, screw it. Seems Jaeger ramps up in memory usage pretty quickly. is just too memory hungry, and probably does tons of memory caches on top, as it's simply designed for a larger scale.

Time to rewrite that thing in . All it really seems to do (in my setup), is to receive tracing data, throw that into a DB and then retrieve it with some filters for the UI. So should be fairly easy to replicate and can re-use the UI.

Sign in to participate in the conversation
Mastodon for Tech Folks

This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!