Skip to content

fix(ops): probe registry before bump image rollout#37

Merged
kadyapam merged 1 commit intomainfrom
kadyapam/bump-image-ghcr-probe
May 5, 2026
Merged

fix(ops): probe registry before bump image rollout#37
kadyapam merged 1 commit intomainfrom
kadyapam/bump-image-ghcr-probe

Conversation

@kadyapam
Copy link
Copy Markdown
Contributor

@kadyapam kadyapam commented May 5, 2026

Summary

  • add a pre-rollout registry manifest probe to automation/agents/noetl/lifecycle/bump_image.yaml
  • retries GHCR/Quay image availability before mutating Kubernetes deployments
  • adds workload knobs for probe attempts and sleep seconds
  • keeps a defensive probe inside bump_components because failed steps can still be followed by the next arc in current workflow semantics

Context

This addresses the v2.35.9 deploy race captured in /Volumes/X10/projects/noetl/ai-meta/bridge/outbox/20260505-191122-v2358-regression-and-fix.result.json, where the GitHub release tag was visible before the GHCR image was pullable and Kubernetes fell into image-pull rollout failure.

Validation

  • Catalog register: automation/agents/noetl/lifecycle/bump_image version 8
  • Negative test: fake tag v9.99.99-fake failed cleanly in the registry probe before mutation, execution 620331512323375641; noetl-server remained on ghcr.io/noetl/noetl:v2.35.9
  • Positive test: v2.35.9 passed probe immediately and all components reported unchanged, execution 620331690296083052

@kadyapam kadyapam force-pushed the kadyapam/bump-image-ghcr-probe branch from e2798db to 8dd448c Compare May 5, 2026 21:27
@kadyapam kadyapam merged commit 827fb86 into main May 5, 2026
@kadyapam kadyapam deleted the kadyapam/bump-image-ghcr-probe branch May 5, 2026 21:29
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