Helm
v8s shows every Helm release in your selected clusters, lets you inspect history and diff revisions, and can roll back, uninstall, or retry a release. It does this with the Helm SDK as you, so hooks, waits, and history pruning behave like the helm CLI. You do not need helm installed. Helm needs no server components, so the Helm section is always present.
Releases table
Releases from all selected clusters are in one table: name, namespace, revision, status, updated, and managed by (a link to the Flux HelmRelease when Flux manages it).
| Status | Tone |
|---|---|
deployed | success |
failed | danger |
pending-install, pending-upgrade, pending-rollback | warning |
superseded | info |
uninstalling | neutral |
Helm keeps each revision in a Secret, and v8s lists releases from the Secret metadata only, so a cluster with hundreds of releases and thousands of revisions loads fast and never downloads the large chart payloads just to show a list. The chart, chart version, and app version columns fill in for the rows you can see, because they live inside the payload. Sorting or filtering by chart decodes the rest in the background, with a progress indicator.
You can see releases if you can list Secrets in the namespace; get is needed only to open one.
Release detail
The release drawer has six tabs:
- Overview. Description, last updated, chart and app version, and the release's hooks (name, kind, events, weight).
- History. Every revision and its status. Pick one revision to roll back to, or tick two to compare.
- Values. Switch between the values you supplied and the computed values.
- Manifest. The rendered manifest.
- Notes. The chart's notes.
- Diff. Pick any two revisions and see the difference in values and in the manifest. The diff is computed in the agent.
Actions
| Action | What it does |
|---|---|
| Roll back | Roll back to a revision you choose. Options: wait, timeout, clean up on failure, and force (behind a confirm) |
| Uninstall | Remove the release. Options: keep history, wait, timeout, cascade policy. Type to confirm in production groups |
| Retry | For a failed or stuck pending-* release, re-run the last upgrade with the chart and values stored in the release. No chart repository access is needed |
For a stuck pending-* release, Retry offers Mark failed (on by default), which is what people do by hand today so the release can be retried.
Each action is a job, so a reload or a dropped connection does not lose track of it, and it keeps running for a few minutes if you close the window.
Flux-managed releases
If a release carries Flux's labels (helm.toolkit.fluxcd.io/name and namespace), Flux owns it, and v8s guards you:
- Roll back and Retry become HelmRelease operations: reconcile, force, or reset failures.
- Uninstall warns that Flux will install the release again unless you suspend or delete the HelmRelease first.
- The direct Helm actions are still there, behind an "I know, do it anyway" confirmation.
See Flux.
Problems from Helm
| Class | Meaning | Default |
|---|---|---|
helm.failed | The latest revision is failed | On |
helm.stuck-pending | A release has been pending-* for more than 15 minutes | On |
See Problems.
Not included
Upgrading a release with edited values is not part of the first release. Use your normal deployment path (Flux, CI, or helm upgrade).