Skip to content

Commit 27e0aa2

Browse files
committed
docs(backups): say who applies a bucket lifecycle rule
The notes that send a tenant to a bucket lifecycle rule for a backup object Cozystack leaves behind read as if some policy were already in place. None is: the rule has to be set on the bucket, and it is applied only by an object store that implements lifecycle, which the bundled SeaweedFS now does once a day while lifecycle.enabled is on. Assisted-by: LLM Signed-off-by: Alexey Artamonov <[email protected]>
1 parent f36d864 commit 27e0aa2

3 files changed

Lines changed: 6 additions & 11 deletions

File tree

‎examples/backups/kafka/92-scenario-user-restore.md‎

Lines changed: 1 addition & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -71,9 +71,6 @@ The script:
7171
had already replayed. The Pod log names each topic before its replay and
7272
again once it landed; empty or delete every topic it shows as started, not
7373
only the one the refusal names, then re-run.
74-
- The tarball in object storage is not managed by the Cozystack `Backup`
75-
lifecycle: deleting the `Backup` CR removes the reference but leaves the S3
76-
object in place. Tenants who need to reclaim space should delete the object
77-
(or set a bucket lifecycle policy).
74+
- The tarball in object storage is not managed by the Cozystack `Backup` lifecycle: deleting the `Backup` CR removes the reference but leaves the S3 object in place. Tenants who need to reclaim space should delete the object, or set a lifecycle rule on the bucket. The Cozystack SeaweedFS applies such rules once a day while its `lifecycle.enabled` is on, which is the default; another object store applies them only if it implements bucket lifecycle.
7875
- Consumer-group offsets, topic configs, ACLs and users are not part of the
7976
backup; consumers start from the restored topic's fresh offsets.

‎examples/backups/nats/92-scenario-user-restore.md‎

Lines changed: 1 addition & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -55,9 +55,6 @@ The script:
5555

5656
- Both flows rely on the strategy template branching on `.Mode`. See
5757
`01-create-strategy.sh`.
58-
- The tarball in object storage is not managed by the Cozystack `Backup`
59-
lifecycle: deleting the `Backup` CR removes the reference but leaves the S3
60-
object in place. Tenants who need to reclaim space should delete the object
61-
(or set a bucket lifecycle policy).
58+
- The tarball in object storage is not managed by the Cozystack `Backup` lifecycle: deleting the `Backup` CR removes the reference but leaves the S3 object in place. Tenants who need to reclaim space should delete the object, or set a lifecycle rule on the bucket. The Cozystack SeaweedFS applies such rules once a day while its `lifecycle.enabled` is on, which is the default; another object store applies them only if it implements bucket lifecycle.
6259
- `nats stream restore` recreates the stream's configuration from the snapshot;
6360
consumers are restored as captured at backup time.

‎internal/backupcontroller/psmdbstrategy_controller.go‎

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1376,9 +1376,10 @@ func (r *BackupJobReconciler) createMongoDBBackupArtifact(
13761376
// annotation (honoured before any apiserver read, so it still frees a Backup
13771377
// whose operator CR has become unreadable) and a going-away namespace (the psmdb
13781378
// delete-backup exec cannot run there) each make the driver strip its own
1379-
// finalizer and release, leaving the object for a bucket lifecycle policy. This
1380-
// bounds only what the driver contributes: a Get that fails with something other
1381-
// than NotFound/NoMatchError still requeues (below), and the psmdb operator's own
1379+
// finalizer and release, leaving the object to a lifecycle rule on the bucket,
1380+
// if one is set, or to manual cleanup. This bounds only what the driver
1381+
// contributes: a Get that fails with something other than
1382+
// NotFound/NoMatchError still requeues (below), and the psmdb operator's own
13821383
// release-lock finalizer keeps the CR Terminating while the operator is down,
13831384
// whatever the driver strips. A down operator in a live namespace is likewise
13841385
// not auto-released: the Backup stays Terminating as a visible signal until an

0 commit comments

Comments
 (0)