-
Notifications
You must be signed in to change notification settings - Fork 559
Description
🔖 Feature description
A new UI button in Devtron that allows users to perform a rollout restart of any selected deployment directly from the Deployments page.
This button will trigger a restart of the pods in the deployment (similar to kubectl rollout restart deployment/), without needing to manually scale or edit the deployment manifest.
🎤 Pitch / Usecases
This feature simplifies common operational tasks and enhances the user experience by reducing the need for manual CLI commands or external tooling. It is especially useful in scenarios such as:
- When environment variables or config maps have been updated externally and a restart is required to pick up changes.
- When diagnosing transient issues where pod recycling is needed without triggering a full redeploy.
- When users want a quick restart of a deployment due to performance degradation, memory leaks, or hung processes.
- Enables non-CLI users and less-experienced team members to perform safe restarts without direct kubectl access.
🔄️ Alternative
An alternative to introducing a dedicated restart button is to rely solely on CLI-based rollout commands (e.g., kubectl rollout restart deployment/<name>) or require users to edit and re-apply manifests or pipeline configurations to trigger a restart.
Another option is to extend RBAC in Devtron so that users can perform rollout restarts via kubectl (or an external script) without UI support; however, this still requires elevated permissions or manual processes.
There is also an existing open issue (#6749) proposing RBAC support specifically for rollout restart operations per environment, which would allow fine-grained permission control but does not provide a direct UI button for users to click within Devtron. Integrating the restart button together with the RBAC support proposed in issue #6749 would offer both usability and secure access control.
👀 Have you spent some time to check if this issue has been raised before?
- I checked and didn't find similar issue
🏢 Have you read the Code of Conduct?
- I have read the Code of Conduct