Skip to content

chore: chart 0.3.1 / appVersion 0.2.1 (JoinExtension priority fix images) - #14

Merged
fupelaqu merged 4 commits into
mainfrom
chore/bump-appversion-0.2.1
Jul 28, 2026
Merged

chore: chart 0.3.1 / appVersion 0.2.1 (JoinExtension priority fix images)#14
fupelaqu merged 4 commits into
mainfrom
chore/bump-appversion-0.2.1

Conversation

@fupelaqu

Copy link
Copy Markdown
Contributor

Companion to SOFTNETWORK-APP/softclient4es-arrow#125 and the 0.2.1 release train.

  • appVersion 0.2.0 → 0.2.1: default federation + per-ES sidecar image tags now deploy the images carrying the JoinExtension priority fix (the collision that silently broke JOINs wherever the community quota extension shared the classpath).
  • Chart version 0.3.0 → 0.3.1 (SemVer bump, changelog comment in Chart.yaml).
  • All 7 golden fixtures regenerated via the same helm template invocations the CI gate uses — the diff is exclusively app.kubernetes.io/version, helm.sh/chart labels and image tags. helm lint clean, default golden re-verified byte-identical.

⚠️ Merge only after the 0.2.1 Docker images (softclient4es-federation, softclient4es{6,7,8,9}-arrow-flight-sql) are published to docker.io/softnetwork — the chart's default tags will 404 (ImagePullBackOff) until then.

Release train: arrow#125 → arrow-extensions 0.2.1 + 0.2.1 images → SoftClient4ES#156 → web#28 → jdbc#19 → this.

🤖 Generated with Claude Code

fupelaqu and others added 4 commits July 27, 2026 20:22
…ority fix

appVersion 0.2.0 → 0.2.1 so the federation server and per-ES sidecar
images carry the arrow-extensions fix (softclient4es-arrow#125: the
quota extension was swallowing JOIN queries at equal priority). Chart
version 0.3.0 → 0.3.1; all 7 golden fixtures regenerated with
helm template (version labels + image tags only). helm lint clean.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The live install jobs pull/kind-load the DockerHub images at IMAGE_TAG so the
chart's appVersion-defaulted image refs resolve to the pre-loaded images.
Chart.yaml moved to appVersion 0.2.1 but IMAGE_TAG stayed at 0.2.0, so every
install job rendered :0.2.1 refs while only :0.2.0 was loaded — ImagePullBackOff.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The ES 8/9 images (~1.4 GB) repeatedly failed in-cluster containerd pulls from
docker.elastic.co on GitHub runners (ImagePullBackOff within the 300s wait) —
two runs in a row, while the smaller ES 6/7 images pulled fine. Pull once on
the host and kind-load instead, mirroring what the workflow already does for
the federation/sidecar images; the tagged image + IfNotPresent then never pulls.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…sters

The pre-pull + kind-load of the ES images tripled their peak footprint (host
docker copy + docker-save tarball in /tmp + containerd copy in the node) and
blew the ~14 GB free on the default runner: 'no space left on device' in every
Deploy ES step. Drop dotnet/Android/GHC/CodeQL first — none are used here.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@fupelaqu
fupelaqu merged commit 6c47577 into main Jul 28, 2026
13 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant