DM-47389: app metrics

[fajpunk]

Publish app metrics events.

Phalanx support

[rra]

I'm not seeing metrics for the lab spawning process, just for notebook execution. Can we get those as well? I think those are going to be really important. (I don't care about metrics for login and certainly not about the intentional pauses, or for things like creating a web socket.)

[fajpunk]

I'm not seeing metrics for the lab spawning process, just for notebook execution. Can we get those as well? I think those are going to be really important. (I don't care about metrics for login and certainly not about the intentional pauses, or for things like creating a web socket.)

Yeah, good call, this is probably one of the most useful events we can have! I added NubladoSpawnLab and NubladoDeleteLab events.

[fajpunk]

I'll bring up in co-work whether we want to replace the current timing stuff with sampled Sentry tracing. If we do decide to do that, we may not want some of these as analytical metrics. Notebook cell executions for example, I don't see the value in keeping that data forever, but I do see the value in keeping it short-term to debug why an entire notebook is suddenly slow

[rra]

This looks good, thanks! I'm fairly sure that there are ways to make some of this code simpler, or at least with somewhat sharper separation of concerns, but I think that's an orthogonal refactoring to the work you're doing, and as you say we should probably wait until we've integrated Sentry and then see what the remaining complexity looks like.