Uploaded image for project: 'OpenShift Pipelines'
  1. OpenShift Pipelines
  2. SRVKP-2622

Reconciler and Controller based pruner for better performance - Tech Preview

XMLWordPrintable

    • Reconciler and Controller based pruner for better performance - Tech Preview
    • 31
    • False
    • Hide

      None

      Show
      None
    • False
    • In Progress
    • SECFLOWOTL-126 - Ship Reconciler and Controller based consistent pruner for all
    • 12% To Do, 0% In Progress, 88% Done
    • Hide
      Release Notes – Pruner Component - TechPreview

      The TechPreview version of the Pruner component helps automatically clean up PipelineRun and TaskRun resources in Tekton.

      Features:
      1. Time-Based Pruning (TTL): Automatically deletes completed PipelineRuns and TaskRuns after a set time Controlled using the ttlSecondsAfterFinished setting.

      2. History-Based Pruning: Keeps a limited number of successful and failed runs. Configured using:
      successfulHistoryLimit
      failedHistoryLimit
      historyLimit (used when the above are not set)

      3. Flexible Configuration Levels
      Pruning settings can be applied at different levels:
      Global level – applies to all namespaces in cluster except the once with prefix kube- and openshfft-(Tech Preview)
      Namespace level – applies to all resources in a namespace (Tech Preview)
      Resource Group level – by pipeline/task name, label, or annotation (Coming Soon)
      Resource level – specific to individual resources (Coming Soon)

      Note: In this release, only Global and Namespace-level configurations are available. The other levels will be added in the next version.
      Show
      Release Notes – Pruner Component - TechPreview The TechPreview version of the Pruner component helps automatically clean up PipelineRun and TaskRun resources in Tekton. Features: 1. Time-Based Pruning (TTL): Automatically deletes completed PipelineRuns and TaskRuns after a set time Controlled using the ttlSecondsAfterFinished setting. 2. History-Based Pruning: Keeps a limited number of successful and failed runs. Configured using: successfulHistoryLimit failedHistoryLimit historyLimit (used when the above are not set) 3. Flexible Configuration Levels Pruning settings can be applied at different levels: Global level – applies to all namespaces in cluster except the once with prefix kube- and openshfft-(Tech Preview) Namespace level – applies to all resources in a namespace (Tech Preview) Resource Group level – by pipeline/task name, label, or annotation (Coming Soon) Resource level – specific to individual resources (Coming Soon) Note: In this release, only Global and Namespace-level configurations are available. The other levels will be added in the next version.
    • Technology Preview
    • Proposed
    • 21

      Epic Goal

      • R&D on pruner based on its own reconciler to have better overall performance.
      • We can also try if we can remove cronjob and perform pruning from reconciler itself (if we are not over killing operator)

      Why is this important?

      • With initial implementation there were n number of containers created for n namespaces, which was resource heavy and made the overall cluster slow.
      • This got better with fix  (https://github.com/tektoncd/operator/pull/805) where we spawn one container to run all the pruning commands.
      • We need to figure out if there are other way to make it better.

      Acceptance Criteria (Mandatory)

      • Pruning based on reconciler if its faster and handle complex configurations.

       

      Previous Story : https://issues.redhat.com/browse/SRVKP-2159

              rh-ee-anataraj Anitha Natarajan
              pradkuma PRADEEP KUMAR (Inactive)
              Anitha Natarajan, Pramod Bindal, Puneet Punamiya
              Sri Vignesh Selvan Sri Vignesh Selvan
              Votes:
              2 Vote for this issue
              Watchers:
              15 Start watching this issue

                Created:
                Updated:
                Resolved: