[GH-ISSUE #441] Support readinessProbe/livenessProbe/startupProbe in the pipelines chart #136

Open
opened 2026-10-18 21:06:33 +00:00 by bsakaguchi · 0 comments
Owner

Originally created by @andriimazurslalom on GitHub (Sep 16, 2026).
Original GitHub issue: https://github.com/open-webui/helm-charts/issues/441

We are using the pipelines chart and noticed there is no way to set readinessProbe, livenessProbe or startupProbe. We checked values.yaml and templates/deployment.yaml in the newest version (open-webui-16.5.0) and this option is not there.

This is a problem for us because our pipelines need some time to start (they import custom pipeline files, and some of them also install dependencies). Without a readiness probe, Kubernetes thinks the pod is ready right after the container process starts, not after the app really finished loading. So during a rolling update, the old pod can get removed before the new one is actually able to answer requests, and this can cause some requests to fail even though the deployment shows as successful.

I could offer the PR with optional readinessProbe, livenessProbe and startupProbe values (default {}, so nothing changes for people who don't use it), and template them into the container spec, something like:

{{- with .Values.readinessProbe }}
readinessProbe:
  {{- toYaml . | nindent 12 }}
{{- end }}

This way each user can point the probe to whatever health endpoint makes sense for their own pipelines, the chart does not need to know anything about that.

If this is something you would accept, I can create a PR for those changes. Should be a small change and should not affect anyone who is not using it.

Originally created by @andriimazurslalom on GitHub (Sep 16, 2026). Original GitHub issue: https://github.com/open-webui/helm-charts/issues/441 We are using the pipelines chart and noticed there is no way to set readinessProbe, livenessProbe or startupProbe. We checked values.yaml and templates/deployment.yaml in the newest version (open-webui-16.5.0) and this option is not there. This is a problem for us because our pipelines need some time to start (they import custom pipeline files, and some of them also install dependencies). Without a readiness probe, Kubernetes thinks the pod is ready right after the container process starts, not after the app really finished loading. So during a rolling update, the old pod can get removed before the new one is actually able to answer requests, and this can cause some requests to fail even though the deployment shows as successful. I could offer the PR with optional readinessProbe, livenessProbe and startupProbe values (default {}, so nothing changes for people who don't use it), and template them into the container spec, something like: ``` {{- with .Values.readinessProbe }} readinessProbe: {{- toYaml . | nindent 12 }} {{- end }} ``` This way each user can point the probe to whatever health endpoint makes sense for their own pipelines, the chart does not need to know anything about that. If this is something you would accept, I can create a PR for those changes. Should be a small change and should not affect anyone who is not using it.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
open-webui/helm-charts#136
No description provided.