treewide: accept list of sstables in "restore" API

before this change, we enumerate the sstables tracked by the
system.sstables table, and restore them when serving
requests to "storage_service/restore" API. this works fine with
"storage_service/backup" API. but this "restore" API cannot be
used as a drop-in replacement of the rclone based API currently
used by scylla-manager.

in order to fill the gap, in this change:

* add the "prefix" parameter for specifying the shared prefix of
  sstables
* add the "sstables" parameter for specifying the list of  TOC
  components of sstables
* remove the "snapshot" parameter, as we don't encode the prefix
  on scylla's end anymore.
* make the "table" parameter mandatory.

Fixes scylladb/scylladb#20461
Signed-off-by: Kefu Chai <kefu.chai@scylladb.com>

test/nodetool/test_restore.py

@@ -8,26 +8,21 @@ import pytest
 
 from test.nodetool.rest_api_mock import expected_request
 
-@pytest.mark.parametrize("table",
-                         ["cf",
-                          pytest.param("",
-                                       marks=pytest.mark.xfail(
-                                           reason="full keyspace restore not implemented yet"))])
 @pytest.mark.parametrize("nowait,task_state,task_error", [(False, "failed", "error"),
                                                           (False, "done", ""),
                                                           (True, "", "")])
-def test_restore(nodetool, scylla_only, table, nowait, task_state, task_error):
+def test_restore(nodetool, scylla_only, nowait, task_state, task_error):
     endpoint = "s3.us-east-2.amazonaws.com"
     bucket = "bucket-foo"
     keyspace = "ks"
+    table = "cf"
+    prefix = "foo/bar"
 
-    snapshot = "ss"
     params = {"endpoint": endpoint,
               "bucket": bucket,
-              "snapshot": snapshot,
-              "keyspace": keyspace}
-    if table:
-        params["table"] = table
+              "prefix": prefix,
+              "keyspace": keyspace,
+              "table": table}
 
     task_id = "2c4a3e5f"
     start_time = "2024-08-08T14:29:25Z"
@@ -48,25 +43,29 @@ def test_restore(nodetool, scylla_only, table, nowait, task_state, task_error):
         "progress_completed": 1.0,
         "children_ids": []
     }
+    # just generate filenames of TOC components of random sstables
+    sstables = [f"me-{id}-big-TOC.txt" for id in range(12)]
     expected_requests = [
         expected_request(
             "POST",
             "/storage_service/restore",
             params,
+            sstables,
             response=task_id)
     ]
     args = ["restore",
             "--endpoint", endpoint,
             "--bucket", bucket,
-            "--snapshot", snapshot,
-            "--keyspace", keyspace]
-    if table:
-        args.extend(["--table", table])
+            "--prefix", prefix,
+            "--keyspace", keyspace,
+            "--table", table]
     if nowait:
         args.append("--nowait")
+        args.extend(sstables)
         res = nodetool(*args, expected_requests=expected_requests)
         assert task_id in res.stdout
     else:
+        args.extend(sstables)
         # wait for the completion of backup task
         expected_requests.append(
             expected_request(

test/object_store/test_backup.py

@@ -153,6 +153,8 @@ async def test_simple_backup_and_restore(manager: ManagerClient, s3_server):
     orig_res = cql.execute(f"SELECT * FROM {ks}.{cf}")
     orig_rows = { x.name: x.value for x in orig_res }
 
+    toc_names = [entry.name for entry in list_sstables() if entry.name.endswith('TOC.txt')]
+
     prefix = f'{cf}/{snap_name}'
     tid = await manager.api.backup(server.ip_addr, ks, cf, snap_name, s3_server.address, s3_server.bucket_name, prefix)
     status = await manager.api.wait_task(server.ip_addr, tid)
@@ -166,7 +168,7 @@ async def test_simple_backup_and_restore(manager: ManagerClient, s3_server):
     assert not res
 
     print(f'Try to restore')
-    tid = await manager.api.restore(server.ip_addr, ks, cf, snap_name, s3_server.address, s3_server.bucket_name)
+    tid = await manager.api.restore(server.ip_addr, ks, cf, s3_server.address, s3_server.bucket_name, prefix, toc_names)
     status = await manager.api.wait_task(server.ip_addr, tid)
     assert (status is not None) and (status['state'] == 'done')
     print(f'Check that sstables came back')

test/pylib/rest_client.py

@@ -317,14 +317,14 @@ class ScyllaRESTAPIClient():
                   "snapshot": tag}
         return await self.client.post_json(f"/storage_service/backup", host=node_ip, params=params)
 
-    async def restore(self, node_ip: str, ks: str, cf: str, tag: str, dest: str, bucket: str) -> str:
+    async def restore(self, node_ip: str, ks: str, cf: str, dest: str, bucket: str, prefix: str, sstables: list[str]) -> str:
         """Restore keyspace:table from backup"""
         params = {"keyspace": ks,
                   "table": cf,
                   "endpoint": dest,
                   "bucket": bucket,
-                  "snapshot": tag}
-        return await self.client.post_json(f"/storage_service/restore", host=node_ip, params=params)
+                  "prefix": prefix}
+        return await self.client.post_json(f"/storage_service/restore", host=node_ip, params=params, json=sstables)
 
     async def take_snapshot(self, node_ip: str, ks: str, tag: str) -> None:
         """Take keyspace snapshot"""