chore: Add stats print for slot migrations (#4456)

* chore: Add stats print for slot migrations

Fixes #4415

tests/dragonfly/cluster_test.py

@@ -214,6 +214,21 @@ async def get_node_id(connection):
     return id
 
 
+def stop_and_get_restore_log(instance):
+    instance.stop()
+    lines = instance.find_in_logs("RestoreStreamer LSN")
+    assert len(lines) == 1
+    line = lines[0]
+    logging.debug(f"Streamer log line: {line}")
+    return line
+
+
+def extract_int_after_prefix(prefix, line):
+    match = re.search(prefix + "(\\d+)", line)
+    assert match
+    return int(match.group(1))
+
+
 @dfly_args({})
 class TestNotEmulated:
     async def test_cluster_commands_fails_when_not_emulate(self, async_client: aioredis.Redis):
@@ -2035,6 +2050,18 @@ async def test_cluster_migration_huge_container(df_factory: DflyInstanceFactory)
     logging.debug(f"Memory before {mem_before} after {mem_after}")
     assert mem_after < mem_before * 1.1
 
+    line = stop_and_get_restore_log(nodes[0].instance)
+
+    # 'with X commands' - how many breakdowns we used for the keys
+    assert extract_int_after_prefix("with ", line) > 500_000
+
+    assert extract_int_after_prefix("Keys skipped ", line) == 0
+    assert extract_int_after_prefix("buckets skipped ", line) == 0
+    assert extract_int_after_prefix("keys written ", line) > 90
+
+    # We don't send updates during the migration
+    assert extract_int_after_prefix("buckets on_db_update ", line) == 0
+
 
 @dfly_args({"proactor_threads": 2, "cluster_mode": "yes"})
 @pytest.mark.parametrize("chunk_size", [1_000_000, 30])
@@ -2056,7 +2083,6 @@ async def test_cluster_migration_while_seeding(
     nodes[0].slots = [(0, 16383)]
     nodes[1].slots = []
     client0 = nodes[0].client
-    client1 = nodes[1].client
 
     await push_config(json.dumps(generate_config(nodes)), [node.admin_client for node in nodes])
 
@@ -2098,6 +2124,12 @@ async def test_cluster_migration_while_seeding(
     capture = await seeder.capture_fake_redis()
     assert await seeder.compare(capture, instances[1].port)
 
+    line = stop_and_get_restore_log(nodes[0].instance)
+    assert extract_int_after_prefix("Keys skipped ", line) == 0
+    assert extract_int_after_prefix("buckets skipped ", line) > 0
+    assert extract_int_after_prefix("keys written ", line) >= 9_000
+    assert extract_int_after_prefix("buckets on_db_update ", line) > 0
+
 
 def parse_lag(replication_info: str):
     lags = re.findall("lag=([0-9]+)\r\n", replication_info)

tests/dragonfly/instance.py

@@ -380,16 +380,17 @@ class DflyInstance:
             for metric_family in text_string_to_metric_families(data)
         }
 
-    def is_in_logs(self, pattern):
+    def find_in_logs(self, pattern):
         if self.proc is not None:
             raise RuntimeError("Must close server first")
 
+        results = []
         matcher = re.compile(pattern)
         for path in self.log_files:
             for line in open(path):
                 if matcher.search(line):
-                    return True
-        return False
+                    results.append(line)
+        return results
 
     @property
     def rss(self):
@@ -416,7 +417,7 @@ class DflyInstanceFactory:
         args.setdefault("noversion_check", None)
         # MacOs does not set it automatically, so we need to set it manually
         args.setdefault("maxmemory", "8G")
-        vmod = "dragonfly_connection=1,accept_server=1,listener_interface=1,main_service=1,rdb_save=1,replica=1,cluster_family=1,proactor_pool=1,dflycmd=1,snapshot=1"
+        vmod = "dragonfly_connection=1,accept_server=1,listener_interface=1,main_service=1,rdb_save=1,replica=1,cluster_family=1,proactor_pool=1,dflycmd=1,snapshot=1,streamer=1"
         args.setdefault("vmodule", vmod)
         args.setdefault("jsonpathv2")