Open Telemetry

The other day I wrote about my DNS scheduler project. Over the last couple of days, I made some additions to implement the industry standard for observability monitoring Open Telemetry. I tried to keep the setup as OTB as possible, and now I’m getting fun water falls in GCP that look like this below.

I’m not going to pretend to understand every last detail of setup, but the value here is huge: tracking a single request through each part of the code will save a ton of trouble if things start breaking. In its current state every request is traced. If new logic is added, it’s a simple #[tracing::instrument(skip(self), err)] or similar macro away.