[Flaky Test] ☂ Latest Flaky Tests

[timuthy]

How to categorize this issue?

/area testing
/kind flake

Which test(s)/suite(s) are flaking:
We've been experiencing failing test runs in this repo lately and many of them are presumably related to undiscovered flakes. This umbrella issue is intended to organize the work and investments done to eliminate those.

For Feb. 11 - 13, 2025 the following failure counts (based on https://prow.gardener.cloud/?repo=gardener%2Fgardener&job=pull-gardener-*&state=failure) which could give us a hint about where to start:

pull-gardener-e2e-kind-ha-multi-zone: 51
pull-gardener-e2e-kind-operator-seed: 40
pull-gardener-e2e-kind-ha-single-zone: 36
pull-gardener-e2e-kind-upgrade: 23
pull-gardener-e2e-kind-migration-ha-single-zone: 23
pull-gardener-e2e-kind-migration: 33
pull-gardener-e2e-kind-ha-multi-zone-upgrade: 22
pull-gardener-e2e-kind-ipv6: 19
pull-gardener-e2e-kind-ha-single-zone-upgrade: 16
pull-gardener-e2e-kind: 13
pull-gardener-e2e-kind-gardenadm: 4
pull-gardener-e2e-kind-ipv6-release-v1-112: 2
pull-gardener-e2e-kind-operator-release-v1-112: 2
pull-gardener-e2e-kind-ha-multi-zone-upgrade-release-v1-112: 1
----
pull-gardener-integration: 17

Apart from the sub-issues below, the following e2e/integration test failures are not yet fixed:

• Skaffold fails to fetch base image (https://cgr.dev/...) (Slack thread, example)
• Node image pulls stuff from internet in every execution (example, example, example) -> we could move the image build to ci-infra similar to this
• @ScheererJ: Gardenlet/Seed integration tests flake because "waiting for CRDs to become healthy" takes too long (example, PR)

[timuthy]

/assign @rfranzke @ialidzhikov @acumino @shafeeqes @marc1404 @tobschli @timuthy

[timuthy]

/assign @timebertt
🙂

[timebertt]

/assign @timebertt 🙂

👍
I will start by working on #11379.

[timuthy]

FYI: Added 74bd4d9 in an unrelated PR to improve the test output.

[rfranzke]

Since #11128 got merged, the integration tests are again flaking more often: https://prow.gardener.cloud/?job=pull-gardener-integration&state=failure (was assumed to have improved after #11503)

cc @tobschli @ScheererJ