Uploaded image for project: 'OpenShift Pipelines'
  1. OpenShift Pipelines
  2. SRVKP-8500

Upstream release notes for Pipelines

XMLWordPrintable

    • False
    • Hide

      None

      Show
      None
    • False
    • Hide
      1. Before, TaskRun/CustomRun creation would fail immediately on transient mutating-webhook timeouts, causing flakiness on busy clusters. This is now resolved by exponential backoff, a `wait-exponential-backoff` ConfigMap, and gated by `enable-wait-exponential-backoff`.

      ```yaml
      # toggle
      apiVersion: v1
      kind: ConfigMap
      metadata:
        name: feature-flags
        namespace: tekton-pipelines
      data:
        enable-wait-exponential-backoff: "true"
      ```

      2. Before, the controller’s thread count was fixed by the compiled default. This is now resolved by allowing `THREADS_PER_CONTROLLER` to override `threads-per-controller` for tuning concurrency.

      3. Before, multiple controller replicas could land on the same node, reducing HA. This is now improved by adding a preferred pod anti-affinity so replicas spread across nodes.

      4. Before, `podTemplate` fields in `TaskRunSpec` didn’t support param substitution, limiting Matrix/multi-arch patterns. This is now resolved by enabling param substitution on `podTemplate` within `TaskRuns`/`TaskRunSpecs`.

      5. Before, variables in `onError` weren’t supported for `pipeline` v1beta1. This is now resolved by adding variable support in `onError`.

      6. Before, the git resolver’s shell-outs didn’t reliably inherit environment variables, breaking env-driven auth/config. This is now resolved, so git subprocesses inherit the pod env.

      7. Before, the resolvers' deployment could accumulate zombie git processes. This is now resolved by switching to a `tini`-based image/entrypoint for resolvers.

      8. Before, the resolvers defaulted `TEKTON_HUB_API` to the public Hub. This is now changed to no default is set because hub.tekton.dev is being shut down; users should set the env var only if pointing to a self-hosted Hub.

      9. (Breaking) Before, several deprecated metrics were still emitted (e.g., `pipelinerun_count`, `running_taskruns_count`). This is now change and deprecated metrics are removed in favor of the newer names (`pipelinerun_total`, `running_pipelineruns`, etc.). Action: update dashboards/alerts to the replacement metrics.

      10. Before, upgrading Tekton would make it impossible to update metadata (like finalizers) on completed PipelineRuns or TaskRuns due to a upgrade-related spec drift not being resolved. Now, post upgrade users can upgrade the metadata on completed pipelineruns and taskruns.
      Show
      1. Before, TaskRun/CustomRun creation would fail immediately on transient mutating-webhook timeouts, causing flakiness on busy clusters. This is now resolved by exponential backoff, a `wait-exponential-backoff` ConfigMap, and gated by `enable-wait-exponential-backoff`. ```yaml # toggle apiVersion: v1 kind: ConfigMap metadata:   name: feature-flags   namespace: tekton-pipelines data:   enable-wait-exponential-backoff: "true" ``` 2. Before, the controller’s thread count was fixed by the compiled default. This is now resolved by allowing `THREADS_PER_CONTROLLER` to override `threads-per-controller` for tuning concurrency. 3. Before, multiple controller replicas could land on the same node, reducing HA. This is now improved by adding a preferred pod anti-affinity so replicas spread across nodes. 4. Before, `podTemplate` fields in `TaskRunSpec` didn’t support param substitution, limiting Matrix/multi-arch patterns. This is now resolved by enabling param substitution on `podTemplate` within `TaskRuns`/`TaskRunSpecs`. 5. Before, variables in `onError` weren’t supported for `pipeline` v1beta1. This is now resolved by adding variable support in `onError`. 6. Before, the git resolver’s shell-outs didn’t reliably inherit environment variables, breaking env-driven auth/config. This is now resolved, so git subprocesses inherit the pod env. 7. Before, the resolvers' deployment could accumulate zombie git processes. This is now resolved by switching to a `tini`-based image/entrypoint for resolvers. 8. Before, the resolvers defaulted `TEKTON_HUB_API` to the public Hub. This is now changed to no default is set because hub.tekton.dev is being shut down; users should set the env var only if pointing to a self-hosted Hub. 9. (Breaking) Before, several deprecated metrics were still emitted (e.g., `pipelinerun_count`, `running_taskruns_count`). This is now change and deprecated metrics are removed in favor of the newer names (`pipelinerun_total`, `running_pipelineruns`, etc.). Action: update dashboards/alerts to the replacement metrics. 10. Before, upgrading Tekton would make it impossible to update metadata (like finalizers) on completed PipelineRuns or TaskRuns due to a upgrade-related spec drift not being resolved. Now, post upgrade users can upgrade the metadata on completed pipelineruns and taskruns.
    • Unspecified Release Note Type - Unknown
    • In Progress

      Story (Required)

      Any relevant RN from upstream needs to be added to the release notes section in a consolidated manner and docs team will be referring to this instead of google docs moving forward.

      Background (Required)

      <Describes the context or background related to this story>

      Out of scope

      • Drafting the release notes text.
      • Review and approval of release notes.
      • Publication of the release notes.

      Approach (Required)

      • Fetch the list of PRs/issues merged upstream for v1.20 from GitHub (tektoncd/pipelines repo).
      • Identify PRs relevant to OpenShift Pipelines (skip upstream-only changes not consumed downstream, if applicable).
      • Organize contributions into categories (features, enhancements, bug fixes).

      Dependencies

      <Describes what this story depends on. Dependent Stories and EPICs should be linked to the story.>

      Acceptance Criteria (Mandatory)

      RN section is added with a consolidated list of revelant upstream release notes.

      INVEST Checklist

      Dependencies identified

      Blockers noted and expected delivery timelines set

      Design is implementable

      Acceptance criteria agreed upon

      Story estimated

      Legend

      Unknown

      Verified

      Unsatisfied

      Done Checklist

      • Code is completed, reviewed, documented and checked in
      • Unit and integration test automation have been delivered and running cleanly in continuous integration/staging/canary environment
      • Continuous Delivery pipeline(s) is able to proceed with new code included
      • Customer facing documentation, API docs etc. are produced/updated, reviewed and published
      • Acceptance criteria are met

              rh-ee-vbobade Vibhav Bobade
              rh-ee-pbheeman Pavan Mandayam Bheeman
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: