chore(deps): bump deck to v1.21.0

[czeslavo]

No description provided.

[pmalek]

Do we want to include some concrete (regression?) tests for Kong/deck#918 which 1.21 includes? 🤔

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +14.5 🎉

Comparison is base (f9f69d1) 60.3% compared to head (60bf374) 74.8%.

Additional details and impacted files

@@           Coverage Diff            @@
##            main   #4108      +/-   ##
========================================
+ Coverage   60.3%   74.8%   +14.5%     
========================================
  Files        150     150              
  Lines      16745   16745              
========================================
+ Hits       10102   12534    +2432     
+ Misses      6012    3434    -2578     
- Partials     631     777     +146     

see 43 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[czeslavo]

Do we want to include some concrete (regression?) tests for Kong/deck#918 which 1.21 includes? 🤔

Actually, that might be a good idea. Hold on, let me try to come up with something.

[czeslavo]

@pmalek That was a fruitful journey that should allow us to make the upgrade tests more reliable as trying to trigger the regression I discovered some lacks of those:

• feat: all-in-one manifests migrations job improvement #4116 (migrations job was not working even if we removed the old one)
• tests: make upgrade tests more robust #4117 (we weren't waiting for the upgraded replicas to become ready and we weren't testing that we're able to change the configuration after the upgrade)

[pmalek]

@pmalek That was a fruitful journey that should allow us to make the upgrade tests more reliable as trying to trigger the regression I discovered some lacks of those:

• feat: all-in-one manifests migrations job improvement #4116 (migrations job was not working even if we removed the old one)
• tests: make upgrade tests more robust #4117 (we weren't waiting for the upgraded replicas to become ready and we weren't testing that we're able to change the configuration after the upgrade)

I'm glad 😅 ... That we found those now, not e.g. after a release.

🎖️

Let's wait for #4117 to be merged then.

[czeslavo]

@pmalek #4117 is ready for review. It triggers the upgrade condition in which FillIDs alone breaks the controller sync loop (https://github.com/Kong/kubernetes-ingress-controller/actions/runs/5142252037/jobs/9255887769). Once we merge it, we will be able to run E2Es against this branch (the one that bumps deck to v1.21.0) and see that the regression is fixed after the bump.

[team-k8s-bot]

E2E (targeted) tests were started at https://github.com/Kong/kubernetes-ingress-controller/actions/runs/5145103305

[czeslavo]

Two e2e don't pass

• run / gke (v1.25.5, TestDeployAndUpgradeAllInOnePostgres_FeatureGates$)
• run / gke (v1.25.5, TestDeployAllInOneDBLESSGateway$)

TestDeployAllInOneDBLESSGateway looks like a regular flake. For TestDeployAndUpgradeAllInOnePostgres_FeatureGates I'm takin a deeper look 😓

[czeslavo]

@programmer04 Ahhh, it's again the problem with GKE E2E tests running against the nightly image 🤦 It will be fixed once we merge it.

[programmer04]

Yeah, it's good to go. 🟢
Btw. we should fix it, let's prfioritize Kong/kubernetes-testing-framework#587 or at least to some hack for GKE's e2e and triggered with label on PR to use image build from this PR @czeslavo