A key made by mistake on a server that is being restored left no way back: key:import and the
passphrase route refused because a key existed, and nothing removed it. As long as the key never
protected anything (no location created or adopted with it, no location or run in the database),
occ backup:key:import --replace-unused-key and occ backup:target:add --replace-unused-key replace
it, after the new key was opened. A key that was used is never replaced, also after its location
was removed. The refusal names the option, and all administrators are notified of a replacement.
On a server without a backup key, backup:target:add finds the backups at
the location and takes their key from its config with one of the old
server's passphrases. Backups then wait until a new recovery kit is
confirmed. The deletion delay and the repository id are now stored on the
location, sealed with the key, so the delay survives a restore without a
kit and --repair-config can write a lost config again from the kit.
The key page warns against creating a new key when restoring a server.
New test: tests/integration/recovery-scenarios.sh, and a no-kit mode in
full-restore-roundtrip.sh.
Backups now record each table's structure in the dump manifest. The full
restore creates tables that do not exist on the new instance (left behind
by apps that were removed or are switched off), lists all problems at
once, and no longer stops on apps that are missing. Foreign keys are not
checked while rows are loaded in name order, and checked again before the
restore commits (PostgreSQL: dropped and added back in the transaction).
--install-apps installs or enables the apps that were enabled at backup
time; --skip-missing-tables leaves out tables of older backups whose
structure is unknown. nextcloud-aio is never required.
backup:target:add without a key points to backup:key:import, and says that
connecting can take a while. Removing a location drops its cached index;
a migration removes the cache of locations that are gone.
The full-restore test now also runs on PostgreSQL with Mail (foreign keys)
and a disabled app.
RepositoryImportSource::copyToFolder() called putContent() on the generic
return type of Node::get(), which doesn't declare that method -- only
File does. Existing-file overwrite during a per-user restore would have
fatally errored at runtime; added an instanceof check and a resource leak
fix (finally-close) while there. Also guarded getFileAsStream() against
fopen() returning false.
Everything else Psalm (strict level) currently flags is mechanical
(missing `final` on leaf classes -- applied everywhere safe) or pre-existing
array-shape/type gaps, captured in psalm-baseline.xml so CI fails only on
new issues from here on, not the whole backlog at once.
- migration: ngb_targets, ngb_runs, ngb_snapshots
- KeyService: master key stored encrypted with the instance secret
(unattended backups), passphrase-wrapped copy in every repository and
in the recovery kit, recorded admin confirmation per key fingerprint
- TargetService: locations built from files_external backends without a
mount; files_external is loaded through IAppManager::loadApp() and its
enabled state is left alone; credentials encrypted with ICrypto;
existing repositories are adopted only if made with this key, and a
repository cannot be registered twice
- Repository: initWithKey/openWithKey with an encrypted key check
- occ: backup:key:init|kit|confirm, backup:target:backends|add|list|remove
dillard