-
Task
-
Resolution: Unresolved
-
Undefined
-
None
-
None
-
None
-
False
-
-
False
Request type: Service Request
Reason of urgency: This is their largest cluster, running builds for many Red Hat products. The number of TaskRuns and PipelineRuns is growing uncontrollably, causing Tekton Pipelines to slow down and impacting overall cluster performance.
Request info
Team: (customer) <redacted due to SRVKP-9147>
Details:
Support Request from @disharma Summary: Is this a known issue that the tekton pruner fails with error: exec /bin/sh: argument list too long if there are numerous namespaces(around 1900) in the cluster. The issue appears to occur because the pruner CronJob’s container definition includes a very long command-line argument list — it enumerates all namespaces/projects as arguments to the pruning command. When the number of namespaces grows large, the container fails immediately with the above error. Customer case: Pipelines Version: 4.18 Workaround: NA Steps to reproduce: - Create around 1900 NS and create pipelines in those NS. - Configure tekton pruner as below: ``` pruner: disabled: false keep-since: 4320 resources: - taskrun - pipelinerun schedule: 30 * * * * ``` Additional Details: Attaching the related logs.
- links to