Add server name label to custom metrics

Author: tulir

synapse/handlers/pagination.py

@@ -32,6 +32,7 @@
 from synapse.events.utils import SerializeEventConfig
 from synapse.handlers.worker_lock import NEW_EVENT_DURING_PURGE_LOCK_NAME
 from synapse.logging.opentracing import trace
+from synapse.metrics import SERVER_NAME_LABEL
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.rest.admin._base import assert_user_is_admin
 from synapse.streams.config import PaginationConfig
@@ -54,7 +55,11 @@
 
 logger = logging.getLogger(__name__)
 
-purge_time = Histogram("room_purge_time", "Time taken to purge rooms (sec)")
+purge_time = Histogram(
+    "room_purge_time",
+    "Time taken to purge rooms (sec)",
+    labelnames=[SERVER_NAME_LABEL],
+)
 
 # How many single event gaps we tolerate returning in a `/messages` response before we
 # backfill and try to fill in the history. This is an arbitrarily picked number so feel
@@ -416,7 +421,9 @@ async def purge_room(
             await self._storage_controllers.purge_events.purge_room(room_id)
 
         purge_end = time.time()
-        purge_time.observe(purge_end - purge_start)
+        purge_time.labels(**{SERVER_NAME_LABEL: self.server_name}).observe(
+            purge_end - purge_start
+        )
         logger.info("purge complete for room_id %s", room_id)
 
     @trace

synapse/handlers/room.py

@@ -75,6 +75,7 @@
 from synapse.events.snapshot import UnpersistedEventContext
 from synapse.events.utils import copy_and_fixup_power_levels_contents
 from synapse.handlers.relations import BundledAggregations
+from synapse.metrics import SERVER_NAME_LABEL
 from synapse.rest.admin._base import assert_user_is_admin
 from synapse.streams import EventSource
 from synapse.types import (
@@ -112,14 +113,20 @@
 FIVE_MINUTES_IN_MS = 5 * 60 * 1000
 
 
-shutdown_time = Histogram("room_shutdown_time", "Time taken to shutdown rooms (sec)")
+shutdown_time = Histogram(
+    "room_shutdown_time",
+    "Time taken to shutdown rooms (sec)",
+    labelnames=[SERVER_NAME_LABEL],
+)
 shutdown_kick_count = Histogram(
     "room_shutdown_kick_count",
     "Number of users successfully kicked while shutting down a room",
+    labelnames=[SERVER_NAME_LABEL],
 )
 shutdown_failed_kick_count = Histogram(
     "room_shutdown_failed_kick_count",
     "Number of users that were failed to be kicked while shutting down a room",
+    labelnames=[SERVER_NAME_LABEL],
 )
 
 
@@ -2457,8 +2464,14 @@ async def shutdown_room(
             result["local_aliases"] = []
 
         shutdown_end = time.time()
-        shutdown_kick_count.observe(len(result["kicked_users"]))
-        shutdown_failed_kick_count.observe(len(result["failed_to_kick_users"]))
-        shutdown_time.observe(shutdown_end - shutdown_start)
+        shutdown_kick_count.labels(**{SERVER_NAME_LABEL: self.hs.hostname}).observe(
+            len(result["kicked_users"])
+        )
+        shutdown_failed_kick_count.labels(
+            **{SERVER_NAME_LABEL: self.hs.hostname}
+        ).observe(len(result["failed_to_kick_users"]))
+        shutdown_time.labels(**{SERVER_NAME_LABEL: self.hs.hostname}).observe(
+            shutdown_end - shutdown_start
+        )
 
         return result

synapse/rest/admin/rooms.py

@@ -46,6 +46,7 @@
     parse_string,
 )
 from synapse.http.site import SynapseRequest
+from synapse.metrics import SERVER_NAME_LABEL
 from synapse.rest.admin._base import (
     admin_patterns,
     assert_requester_is_admin,
@@ -67,6 +68,7 @@
 delete_time = Histogram(
     "admin_room_delete_time",
     "Time taken to delete rooms via the admin API (sec)",
+    labelnames=[SERVER_NAME_LABEL],
 )
 
 
@@ -356,7 +358,9 @@ async def on_DELETE(
         )
         end = time.time()
         logger.info(f"[admin/rooms] deleting {room_id} took {end - start} seconds")
-        delete_time.observe(end - start)
+        delete_time.labels(**{SERVER_NAME_LABEL: request.our_server_name}).observe(
+            end - start
+        )
         return response
 
     async def _delete_room(