Walk datasets rather than directory for backups

playbooks/filesystem/yggdrasil/etc/systemd/system/restic-service-data.service.j2

@@ -8,4 +8,4 @@ Environment=TZ=UTC
 Environment=RESTIC_CACHE_DIR=/var/cache/restic
 Environment=RESTIC_PASSWORD_FILE=/etc/restic.password
 EnvironmentFile=/etc/scaleway.keys
-ExecStart=/usr/local/sbin/restic-service-data --data-root /var/lib/{{ ansible_hostname }}/data --bucket-endpoint {{ scw_bucket_endpoint }}
+ExecStart=/usr/local/sbin/restic-service-data --root-dataset rpool/var/lib/{{ ansible_hostname }}/data --bucket-endpoint {{ scw_bucket_endpoint }}

playbooks/filesystem/yggdrasil/usr/local/sbin/restic-service-data

@@ -5,42 +5,50 @@ import os
 import subprocess
 
 
-def get_service_dataset_paths(data_root):
+def get_service_datasets(root_dataset):
-    return { d: os.path.join(data_root, d) for d in os.listdir(data_root) }
+    zfs_list = subprocess.getoutput(
+        f"zfs list -H -r {root_dataset} -o name,mountpoint"
+    )
+    zfs_list_lines = zfs_list.split('\n')
+    zfs_list_lines_items = map(lambda l: l.split(), zfs_list_lines)
+    return {
+        os.path.basename(dataset): {
+            "dataset": dataset,
+            "mountpoint": mountpoint,
+        } for dataset, mountpoint in zfs_list_lines_items if dataset != root_dataset
+    }
 
-def get_last_daily_snapshot_name(dataset_path):
+def get_last_daily_snapshot(dataset):
-    dataset = ''.join(["rpool", dataset_path])
     snapshots = subprocess.getoutput(
         f"zfs list -t snapshot -H -r {dataset} -o name -s creation"
     )
     daily_snapshots = filter(lambda s: s.endswith("_daily"), snapshots.split('\n'))
-    return list(daily_snapshots)[-1]
+    last_daily_snapshot = list(daily_snapshots)[-1]
+    assert '@' in last_daily_snapshot
+    assert last_daily_snapshot.split('@')[0] == dataset
+    return last_daily_snapshot
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Backup service data using restic")
-    parser.add_argument("--data-root", type=str, required=True,
+    parser.add_argument("--root-dataset", type=str, required=True,
-                        help="Service data root")
+                        help="The root data root whose children are to be backed up")
     parser.add_argument("--bucket-endpoint", type=str, required=True,
-                        help="S3 bucket endpoint")
+                        help="S3 bucket endpoint for the backups")
     args = parser.parse_args()
 
-    snapshots_for_backup = {
+    service_datasets = get_service_datasets(args.root_dataset)
-        service: {
+    for service, properties in service_datasets.items():
-            "dataset_path": service_dataset_path,
+        properties["snapshot"] = get_last_daily_snapshot(properties["dataset"])
-            "snapshot": get_last_daily_snapshot_name(service_dataset_path),
-        } for service, service_dataset_path in get_service_dataset_paths(args.data_root).items()
-    }
 
-    for service, info in snapshots_for_backup.items():
+    for service, properties in service_datasets.items():
-        backup_path = os.path.normpath(
+        mountpoint = properties["mountpoint"]
-            os.path.join("/", "mnt", os.path.relpath(info["dataset_path"], "/"))
+        snapshot = properties["snapshot"]
-        )
+        backup_path = os.path.normpath(os.path.join("/", "mnt", os.path.relpath(mountpoint, "/")))
-        snapshot = info["snapshot"]
         restic_cmd_base = "restic " \
             f"--repo s3:https://{args.bucket_endpoint}/the-nine-worlds---{service} " \
             "--option s3.storage-class=ONEZONE_IA"
 
-        print(f"Backing up {service} : {snapshot}")
+        print(f"Backing up {service} : {snapshot}", flush=True)
 
         subprocess.run(f"zfs clone -o mountpoint={backup_path} {snapshot} rpool/restic",
                        shell=True, check=True)