feat: add monitoring to the docker image

[mathieu-pousse]

As we are now using this docker image in prod, we needed to add some monitoring to understand the overall performances of the project.

Here is the result of what is running in prod for us.

We are able to produce such dashboard to ensure everything is running fine and set alerting on it. This also helps us to see the most time consuming transformation was the SeparateKeyframes that was useless for us and lead me to the previous MR :smile:

BTW, as we are running in k8s, the fact of having metrics exposed will help us to configure an HorizontalPodAutoscaler based on those metrics.

My main concern is about the refactoring I did of the profiling in the project and the way to write the doProfile method. Let me know.

Hope it helps

[mathieu-pousse]

I pushed the fixes

I am wondering how does the CI build the image.

Will it firstly build the new version of @ampproject/toolbox-optimizer then build the image that will pull the freshly created new bundle ?

[samouri]

I am wondering how does the CI build the image.

We actually don't have a good system for this yet. Right now we manually build + push new images. I'll create an issue to integrate image build/release with CD.

[mathieu-pousse]

Did I broke the CI with my PR ? Let me know

[sebastianbenz]

Noticed one problem that was causing the optimizer tests to fail.

[sebastianbenz]

And there seems to be another problem that jest fails to init optimizer-docker/metrics.js. One simple fix is to ignore the optimizer-docker dir during tests by adding this to jest.config.js:

   // An array of regexp pattern strings that are matched against all test paths, matched tests are skipped
-  // testPathIgnorePatterns: [
-  //   "/node_modules/"
-  // ],
+   testPathIgnorePatterns: [
+     "/node_modules/",
+     "<rootDir>/packages/optimizer-docker"
+   ],

[mathieu-pousse]

:tada: