When Dokku runs in Docker with a shared socket, create/chown still use
DOKKU_LIB_ROOT inside the container, but Docker -v flags for storage:exec
and deploy must use DOKKU_LIB_HOST_ROOT so the daemon mounts the host data
volume instead of the real host's /var/lib/dokku.
Closes#9126.
The ephemeral containers that run `app.json` deploy tasks ignored a failure from the `docker-args-deploy` and `docker-args-process-deploy` triggers, so a plugin refusing to describe the container produced a task that ran without the arguments it had asked for, while the very same failure aborted the deploy moments later on the scheduler's own path. Both are now fatal, which the trigger documentation states. Storage no longer reports a scheduler mismatch for an app deploying with another scheduler, since those triggers are fired for every app whatever its scheduler and a k3s app's volumes are mounted by k3s rather than bound into a container here.
An attachment's process type was recorded, reported and used to scope `storage:mount --replace` and `storage:unmount --all`, but nothing read it when the volume was actually mounted, so `storage:mount <app> <entry> --container-dir <path> --process-type web` bound the volume into every process rather than only `web`. Both schedulers now scope a mount to the process type it was declared for, and containers that belong to no process type - a `dokku run` one-off, an `app.json` deploy task, a k3s cron job - receive default-scoped mounts only. A container path can now be claimed by a single mount per process type, and `storage:migrate` drains process-scoped `-v` lines out of `docker-options` rather than only the default-scoped ones.
`storage:create` and `storage:set` accept a `--mode` flag that sets the octal permissions of a docker-local host directory, and `storage:destroy` accepts a `--destroy-host-dir` flag that removes the directory along with its contents. A docker-local entry also honors `--reclaim-policy Delete` at destroy time now, matching how that policy governs a k3s PersistentVolume. Both are limited to the default `/var/lib/dokku/data/storage/<name>` location, the same restriction `--chown` already carries. `storage:set` applies `--chown` and `--mode` to the directory rather than only recording them.
Adds an app-aware `fn-verify-app-image` helper that pulls the deployed image
from the configured registry when it is missing locally, gated on
push-on-release being enabled, honoring per-app `registry:login` credentials.
Every in-tree `verify_image` caller migrates to the wrapper and the bare
helper is marked deprecated. `config:set` now forwards the deployed image tag
to `release-and-deploy` so the release path no longer pins recovery on the
unrecoverable `:latest` tag.
The docker-options DOCKER_OPTIONS_<PHASE>.migrated sentinels and the storage data/storage-registry/migrations/<app> flag files are now stored as per-app properties (`migrated-<phase>` and `legacy-mounts-migrated` respectively), so they survive a property-store backup/restore and align with the existing `migrated-from-files`, `env-migrated`, and `nginx-conf-sigil-migrated` patterns. The new markers are written only when a non-empty legacy source was actually migrated, distinguishing apps that never had legacy state from apps that did and were drained. A one-cycle upgrade-cycle conversion drains any leftover filesystem sentinels into the new properties and removes the files; for docker-options the conversion deliberately runs before the global `migrated-from-files` short-circuit so installs upgrading from the previous release still get their `.migrated` files converted.
`SaveEntry` and `touchMigrationFlag` wrote files via `os.WriteFile` and `os.Create` and never chowned them, so the install-time legacy-mount migration produced root-owned `legacy-*.json` files in `/var/lib/dokku/data/storage-registry/entries/`. The dokku user that runs `ps:rebuild` could not read them and every rebuild on a 0.37.x to 0.38.0 upgrade failed with permission denied. The companion `repairRegistryOwnership` pass in `TriggerInstall` rewrites ownership across the whole registry tree so installs that already ran the buggy code are fixed on the next package upgrade, since the per-app migration flag would otherwise cause `MigrateLegacyMounts` to skip the broken files forever.
The install trigger created /var/lib/dokku/data/storage-registry and its entries / migrations subdirectories as root with mode 0755, so the dokku user that runs storage:create couldn't write entry JSON files or migration flag files into them. Mirrors what PropertySetup does for config/storage by calling SetPermissions on each registry path. This unblocks the bats coverage that was failing at the first storage:create call.
storage:list was calling the storage-list plugn trigger which read -v lines from docker-options. After the install-time migration drains those lines into the attachment store, that source is empty for every migrated app and for any app that only ever used storage:create + storage:mount. The fix moves CommandList to call a new in-process ListAppMountEntries helper that reads attachments directly, surfaces the entry name in JSON output via a new entry_name field, and falls back to the entry name as the host token for k3s entries with no host path so the colon form remains well-formed. The storage-list plugn trigger is kept for back-compat with external callers but emits a deprecation warning and now reads from the same attachment-driven source.
Switches the storage plugin's post-delete and post-app-rename-setup triggers from rewriting the attachment list as empty to removing the per-app property folder outright via PropertyDestroy. This matches the convention other plugins use on app deletion and avoids leaving an empty mounts file behind for an app that no longer exists.
apps:destroy was failing in CI because the post-delete trigger called PropertyListWrite, which tried to mkdir /var/lib/dokku/config/storage/<app> as the dokku user but the parent directory had been created by the install trigger as root. The install trigger now calls PropertySetup("storage") so the per-plugin config root has the correct ownership, and post-delete / post-app-clone / post-app-rename short-circuit when the app has no attachments to begin with so they can't fail on a never-touched property tree. The entry registry and migration flag directories also moved out of config/storage (where they would have collided with the property-store path config/storage/<appName>) into a dedicated data/storage-registry tree.
storage:mount and storage:unmount now accept the named-entry form (--container-dir, --phase, --process-type, --volume-subpath, --volume-readonly, --volume-chown), persisting attachments via the property list. The legacy host:container colon form is preserved for back-compat. Storage now emits its own docker-args-deploy and docker-args-run triggers so docker-local apps continue to receive bind mounts even after their data moves out of docker-options. The install trigger walks every app once per upgrade and converts legacy -v lines into legacy-<hash> entries plus attachments, draining the originals from docker-options last so a partial run leaves the old behavior intact. Lifecycle triggers post-delete, post-app-clone-setup, and post-app-rename-setup keep the attachment store in sync with app state.