[jskrablin]

Take a look at the OTEL (Open Telemetry) tooling and libraries. Or Grafana stack/offering with Prometheus, Tempo and Loki. Centralized logging and service calls/code execution tracing is not exactly new. It is often an afterthought.. and then you get yourself is this kinds of unpleasant situations.

And since you didn't implement correct tooling from the start, your team is even smaller and more limited... because you have little to zero idea on what your services are up to.

As per db instances... you upgrade them one by one. Unless there's some really bad bugs present (security or otherwise) there's no rush in upgrading stuff just because.