search

culibraries / folio

Project management for FOLIO implementation at CU Boulder
https://github.com/orgs/culibraries/projects/12
MIT License
0 stars 0 forks source link

Issue 206 deploy mem and cpu logging #252

Closed steveellis closed 2 years ago

steveellis commented 2 years ago

Closes #206.

After doing quite a bit of research the way we want to do this is have both monitoring and logging flow through CloudWatch and bypass Prometheus entirely. The Prometheus we had with rancher was an enhanced one. The one you get via helm with the default config anyway is very basic and nearly unusable. I'm sure with a lot of config you could get it to where it produces something meaningful but we don't have time for that.

The good news is that piping monitoring metrics directly into CloudWatch is easy to set up and has a great dashboard UX.

This isn't about logging, but from what I see the flow there should also be to pipe logs to CloudWatch and then optionally pipe them from there into ES. In other words CloudWatch is indended for and probably should be our "single pane of glass".

I'm leaving the Prometheus code in this PR since it does work, but I'm removing it from index.ts. Happy to remove it entirely.

The other thing I learned was that there's not much value for us at least for now to customizing the AWS yaml files that do the monitoring deployment. It had the metrics I cared about and a whole lot more. So it's all just a script file and not in pulumi.

steveellis commented 2 years ago

Removed prometheus code. Also think we agreed to keep the yaml files as is for now.