hull purge

Find every release hull has installed anywhere in the cluster, uninstall each one, and optionally delete the namespaces and CRDs hull created.

When to use it

Tearing down a test or CI cluster: remove everything hull put there in one command instead of uninstalling releases one at a time.
Recovering after a node or cluster failure that left releases half-removed — --force deletes storage directly and unsticks Terminating namespaces.

hull purge is the cluster-wide broom that hull uninstall is not: uninstall removes one named release, purge sweeps them all.

What happens

Connects to the cluster and lists every release-storage Secret labelled managedBy=hull (or the legacy owner=hull), reconstructing each release’s name and namespace. Resources hull never installed are never touched.
Also picks up any namespace hull labelled at creation time, even if its releases are already gone.
Applies your scope filters: --ns-prefix narrows to matching namespaces, --exclude-ns protects specific ones. kube-system, kube-public, kube-node-lease, and default are always protected.
Without --yes, prints the plan and stops — nothing changes. With --yes, uninstalls every in-scope release (deleting its stored history), then optionally deletes namespaces (--delete-namespaces) and hull CRDs (--delete-crds).

--yes mutates the cluster and requires a reachable cluster. Running with neither --dry-run nor --yes is refused.

Usage

hull purge [flags]

Flags

Flag	Type	Default	Description
`--dry-run`	—	false	preview only; print what would be removed and exit
`--yes`	—	false	actually run; required when `--dry-run` is not set
`--ns-prefix`	string	—	restrict scope to namespaces beginning with this prefix
`--exclude-ns`	stringArray	—	namespaces to skip (repeatable, comma-separated)
`--delete-namespaces`	—	false	after uninstall, delete every namespace that held a hull release (never `kube-*`/`default`)
`--delete-crds`	—	false	remove hull-installed CRDs (currently `hullreleases.hull.dev`)
`--parallel`	int	4	number of namespaces to purge concurrently
`--ignore-failures`	—	false	keep going past failed uninstalls; report the count at the end
`--force`	—	false	skip graceful uninstall: delete storage Secrets directly, force-delete pods, force-finalize stuck namespaces (use after a node failure)

Persistent flags inherited from `hull`

Flag	Type	Description
`--debug`	—	enable debug output
`--kube-context`	string	Kubernetes context to use
`--kubeconfig`	string	path to kubeconfig file
`-n, --namespace`	string	Kubernetes namespace

Worked example

INPUT — three hull releases live in test namespaces:

hull list --all-namespaces

NAME     NAMESPACE   REVISION   STATUS     UPDATED
api      hull-test   3          deployed   2026-07-18 09:10:22
cache    hull-test   1          deployed   2026-07-18 09:11:04
worker   hull-ci     2          deployed   2026-07-17 22:40:51

Preview first — nothing is touched:

hull purge --dry-run --ns-prefix hull-

found 3 release(s) across 2 namespace(s) in scope

--- DRY RUN ---
  ns/hull-ci
    └ release/worker
  ns/hull-test
    └ release/api
    └ release/cache

(re-run with --yes to actually purge)

Execute — uninstall the releases and drop the namespaces:

hull purge --yes --ns-prefix hull- --delete-namespaces

found 3 release(s) across 2 namespace(s) in scope
  ✓ hull-ci/worker [1/3]
  ✓ hull-test/api [2/3]
  ✓ hull-test/cache [3/3]
  ✓ ns/hull-ci deleted
  ✓ ns/hull-test deleted
purge complete

OUTPUT — the releases and their history are gone:

hull list --all-namespaces

NAME   NAMESPACE   REVISION   STATUS   UPDATED

Each release’s stored revisions are deleted along with it, so hull history and hull rollback have nothing left to act on.