helmfile diff, apply, and sync

Helmfile gives you three core commands for changing a cluster: diff to preview, apply to converge smartly, and sync to converge unconditionally.

helmfile diff shows what would change without touching the cluster. It is your safety check before any real deploy.

The diff command relies on the helm-diff plugin under the hood. Install it once so Helmfile can render a clean before-and-after.

helmfile apply runs a diff first, then upgrades only the releases that actually changed. It skips work when nothing differs.

Because apply checks the diff first, unchanged releases are left alone. That means fewer pointless upgrades and faster, calmer deploys.

helmfile sync runs helm upgrade --install on every release no matter what. Use it when you want a guaranteed full reconcile.

The key difference: apply changes only what differs, while sync upgrades all releases blindly. apply is the everyday default; sync is the big hammer.

Use a selector to act on a subset. Label releases, then pass -l to diff or apply just the ones you mean.

helmfile destroy uninstalls every release defined in the file. It is the clean way to tear down a whole environment at once.

The habit that saves you: run diff, read it carefully, then run apply. Preview first, change second, every single time.

In CI you post the diff on a pull request for review, then run apply on merge. That turns deploys into reviewable, auditable changes.

You want to update only the releases that actually changed, skipping the rest. Which command fits best?

You learned to preview with diff, converge changed releases with apply, and force a full reconcile with sync. You can now run a multi-env stack. 🎯