Juju Team Updates - Week #48 2021

Hi everyone, below you can find the updates from the Canonical Juju Team for the week starting on the 29/11/2021. Have a great weekend!

cc @pedroleaoc @jameinel

This week:

  • Have been working on IAM Instance Profile tear down errors and clean up of instance profiles.
  • Investigating the use of tracing in Juju api.
  • Work on failing gating tests for CAAS
1 Like
1 Like
  • Fix storage hooks lack of env vars to describe the storage the hook relates to #13538 LP1948228
  • Crossport restart package from snapd pebble#87
  • Fix various missing error checks and missing wait mixin for CLI commands with wait pebble#88
  • Add finaliser support to pebble so juju charms have a chance to run teardown logic before the sidecar containers go away pebble#89
  • Continue working on juju teardown for sidecar charms.
1 Like
1 Like
  • Juju 2.9.21 release
  • add 2 new Azure regions (uswest3, swedencentral) to Juju repo (still need to publish YAML)
  • fix for mongo replicaset configuration when Juju HA setup changes
  • fixes to juju backup functionality
    • properly handle backup-dir model config when mongo is running as a snap
    • remove option to store backup in mongo
  • customer support tasks
1 Like
  • Working on improvements to Juju docs around LXD Profiles, juju and charms.
  • ERROR “juju.worker.uniter agent.go:31 resolver loop error: executing operation “fail action 640” for /4: state changing too quickly; try again soon” popping up again in 2.9.18 and 2.8.11+. If you see this, please check out LP 1953077 and help us solve it!
  • Fixing bugs around using charm channels in a bundle with a pre-2.9 juju controller and juju 2.9 client.
1 Like
  • Continue working on spec about Advanced Juju Networking
  • Participate in networking discussion with OSM team
  • Study materials about Juju and OSM project collaboration
1 Like

Worked on

  • restructuring (e.g., Juju and LXD profiles)
  • listing (e.g., docs related to upgrading, configuring, working offline)
  • other routine improvements

Also gave feedback on a new charm building tutorial related to LXD.

Also thought again about how to refine the Juju story from the service consumer perspective (as opposed to the service provider perspective).

1 Like
  • Work on identifying what it would require to better expose an SDK, so others can use the Juju API.
  • Design work on the watcher API so that we can drive it from dqlite.
  • Initial scaffolding tests for adding dqlite to Juju, including HA mode (a lot more to do!)
  • Add metrics to the dependency engine to help operators of Juju see when workers restart. The aim is then to add alerts to the metrics if a runaway happens.
  • Pylibjuju fix for missing watcher ids in API calls.
1 Like
  • Collaborated with Simon on the watch API refactoring and the dqlite scaffolding work.
1 Like

pylibjuju updates:

Big thing for this week is: we finally have all greens in the the CI tests! :confetti_ball: :tada:

  • The charms used in the tests are updated to use the Charmhub. [PR #597]
  • Added support for local type: file resources definitions [PR #590]
  • Implemented debug-log parameters. [PR #595]
  • Updated the schemas for talking to Juju 2.9.19. [PR #594]
  • Added some small fixes/patches [PR #593] [PR juju-qa-jenkins#679]

Coming up:

  • 2.9.5 Release is ready and coming. [PR #599]
2 Likes
  • Finally fixed and landed the device provider ID patch mentioned in prior updates.
  • Ongoing analysis of internal cloud issues.
  • Design meetings for Juju networking and state back-end changes.
  • Patch preventing possible state pool reference leakages.
  • Started work on a patch to make accessing units in state more efficient, by storing a stringified charm URL instead of an actual object reference.
1 Like