Skip to content

Wait for ZooKeeper deployment rollout in the Dubbo ingress CI - #7427

Merged
Aias00 merged 1 commit into
apache:masterfrom
BobSong-dev:fix/k8s-dubbo-zookeeper-readiness
Oct 4, 2026
Merged

Aias00 merged 1 commit into
apache:masterfrom
BobSong-dev:fix/k8s-dubbo-zookeeper-readiness

Conversation

@BobSong-dev

@BobSong-dev BobSong-dev commented Oct 3, 2026 •

Copy link
Copy Markdown
Contributor

Background

The Dubbo ingress setup applies the ZooKeeper Deployment and immediately waits for Pods matching its label. Kubernetes creates those Pods asynchronously, so kubectl wait can fail with no matching resources found before a Pod exists. This happened in the Kubernetes Dubbo job for #7424, before ShenYu deployment or business tests started. The same setup script remains in master.

Changes

  • Wait for the named ZooKeeper Deployment to finish rolling out, with an explicit 120-second timeout. Preserve nonzero failures and stop before subsequent deployments if readiness fails.
  • Add five Bash regression cases covering delayed Pod creation, rollout timeout, Forbidden responses, manifest application failure and image loading failure.
  • Run the regression script only in the Dubbo ingress matrix branch, before installing Kubernetes.

Verification

  • Both scripts passed bash -n; all five Bash regression cases passed in Windows Git Bash and an existing RISC-V Linux container (repeated cases, not ten distinct tests). The same delayed-Pod case failed against the original master script with no matching resources found.
  • Five controlled API checks passed with Windows kubectl v1.24.14, matching the workflow's client version. Earlier v1.36.1 checks remain separate evidence.
  • Five readiness checks passed on an isolated real kind v0.14.0 / Kubernetes v1.21.1 cluster using kubectl v1.24.14 and the repository's ZooKeeper manifest. A temporary Pod quota reproduced the old missing-Pod failure; the production 120-second rollout command remained pending until Pods could be created and then succeeded. Already-ready rollout, actual ZooKeeper server status and bounded failure (2-second negative-case limit) also passed. The existing client/server version-skew warning was retained.
  • Workflow YAML and matrix/step assertions passed; git diff --cached --check passed.
  • Root-only Maven Wrapper validate and apache-rat:check passed: zero Checkstyle violations and zero unapproved/unknown licenses.

@Aias00 Aias00 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed the current head. The Dubbo ingress CI now waits on the ZooKeeper Deployment rollout with a bounded timeout, and the failure modes are covered by the script tests. No blocking issues found.

@Aias00
Aias00 merged commit 9ef717e into apache:master Oct 4, 2026
23 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.

2 participants