flagger
flagger copied to clipboard
Update dependencies to Kubernetes 1.30
Changes:
- Update
k8s.io/*
packages to v0.30.0. - Update
fmt
andtest-fmt
in Makefile to replace use ofgofmt
. - Refactor code generation utility hack and replace deprecated
k8s.io/code-generator
codegen scripts.
Codecov Report
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 34.10%. Comparing base (
e3a529e
) to head (5e6815d
). Report is 18 commits behind head on main.
Additional details and impacted files
@@ Coverage Diff @@
## main #1638 +/- ##
===========================================
- Coverage 56.51% 34.10% -22.42%
===========================================
Files 85 282 +197
Lines 8571 20557 +11986
===========================================
+ Hits 4844 7011 +2167
- Misses 3053 12616 +9563
- Partials 674 930 +256
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Hey 👋 I believe this will trigger https://github.com/fluxcd/flagger/issues/1201 . We should warn somewhere that an upgrade will trigger a rollout of every canary.
As a solution for the previous issue could we change the hash function so that it doesn't change during a lib upgrade? Thanks
The serialisation seems to be same, otherwise the CI would fail in this PR.
Can you point me to the test that is doing it for the deployment spec? I could not find it and doing local tests we did end up with differences using the lib for k8s 1.27 vs 1.30
There is one important addition from K8s 1.28 that could cause the above-mentioned issue: sidecar container support via the new .initContainers[].restartPolicy
field. This is the reason that prompted me to submit the PR in the first place 😄
I've updated the kind node version used in e2e tests. Pinned to v1.29.2 as that's the latest stable version available.
@relu there is a new kind release that supports 1.30, please update the workflow to https://github.com/kubernetes-sigs/kind/releases/tag/v0.23.0