Framework Team Updates - Week #11 2022

Add this week’s updates here. @ppasotti, @pengale, @jnsgruk

  • Working on improvements for Container.can_connect in the harness addressing #691
  • Working on pebble log trimming
  • Finished up pebble file permission PR
  • Addressed review on and improved streaming pull PR - merge imminent hopefully?
  • Various ponderings about charm/relation versioning
  • Made a few doc improvements/edits on discourse about relations.

What software is considered “framework” ? Can this be explained in these threads?

1 Like

He’s referring to the Operator Framework team, those working on the ops library and charming libs.

1 Like
  • Published some scripts for (among others) fast-track packing a .charm while we wait for the official charmcraft functionality (I’ve been using them for a while, but published them only now).
  • Worked some more on the integration tests for loki-k8s-operator charm.
  • Made good progress on the traefik charm, stripped out the whole versioning thing and ensured the tests still pass. Will need some review from Michele at some point soon.
  • Attended most FINOS/legend meetings this week, interesting things going on over there, but right now they’re mostly busy with CI/CD and release setup, not much to contribute from my side.

Last week, I:

  • Did a deep dive into the ops codebase, refreshing my picture of how the Charm and Framework objects get instantiated and passed around.
  • Commented on the github #719, which is the issue that sparked the above investigation. I recommended against refactoring to add import style access to either state.
  • Refactored the Rolling Ops prototype to be more specific and concise about accessing the parent Charm and Framework objects. (see above)
  • Investigated why juju wait does not work on the homebrew or standalone binary builds of Juju (it’s only included in the snap build).
  • Fixed the behavior of the test harness around setting initial app state in begin_with_initial_hooks, addressing gh #726.
  • Helped Pietro step through an issue w/ timing in an intregration test.
  • Spent Friday making a timeboxed attempt at a “workload events” library, as an attempt to fix the issues we ran into above. The idea I initially had won’t work, but I think that the time spent was worthwhile. (Will have more thoughts this week.)