From 1e19ce00bff8d67168d39201cdf9424f7b2f22f6 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 17 Sep 2019 11:41:54 +0100 Subject: Add 'failure_ts' column to 'destinations' table (#6016) Track the time that a server started failing at, for general analysis purposes. --- .../schema/delta/56/destinations_failure_ts.sql | 25 ++++++++++++++++++++++ synapse/storage/transactions.py | 23 ++++++++++++++------ synapse/util/retryutils.py | 16 +++++++++++++- 3 files changed, 56 insertions(+), 8 deletions(-) create mode 100644 synapse/storage/schema/delta/56/destinations_failure_ts.sql (limited to 'synapse') diff --git a/synapse/storage/schema/delta/56/destinations_failure_ts.sql b/synapse/storage/schema/delta/56/destinations_failure_ts.sql new file mode 100644 index 0000000000..f00889290b --- /dev/null +++ b/synapse/storage/schema/delta/56/destinations_failure_ts.sql @@ -0,0 +1,25 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Record the timestamp when a given server started failing + */ +ALTER TABLE destinations ADD failure_ts BIGINT; + +/* as a rough approximation, we assume that the server started failing at + * retry_interval before the last retry + */ +UPDATE destinations SET failure_ts = retry_last_ts - retry_interval + WHERE retry_last_ts > 0; diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index d81ace0ece..289c117396 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -165,7 +165,7 @@ class TransactionStore(SQLBaseStore): txn, table="destinations", keyvalues={"destination": destination}, - retcols=("destination", "retry_last_ts", "retry_interval"), + retcols=("destination", "failure_ts", "retry_last_ts", "retry_interval"), allow_none=True, ) @@ -174,12 +174,15 @@ class TransactionStore(SQLBaseStore): else: return None - def set_destination_retry_timings(self, destination, retry_last_ts, retry_interval): + def set_destination_retry_timings( + self, destination, failure_ts, retry_last_ts, retry_interval + ): """Sets the current retry timings for a given destination. Both timings should be zero if retrying is no longer occuring. Args: destination (str) + failure_ts (int|None) - when the server started failing (ms since epoch) retry_last_ts (int) - time of last retry attempt in unix epoch ms retry_interval (int) - how long until next retry in ms """ @@ -189,12 +192,13 @@ class TransactionStore(SQLBaseStore): "set_destination_retry_timings", self._set_destination_retry_timings, destination, + failure_ts, retry_last_ts, retry_interval, ) def _set_destination_retry_timings( - self, txn, destination, retry_last_ts, retry_interval + self, txn, destination, failure_ts, retry_last_ts, retry_interval ): if self.database_engine.can_native_upsert: @@ -202,9 +206,12 @@ class TransactionStore(SQLBaseStore): # resetting it) or greater than the existing retry interval. sql = """ - INSERT INTO destinations (destination, retry_last_ts, retry_interval) - VALUES (?, ?, ?) + INSERT INTO destinations ( + destination, failure_ts, retry_last_ts, retry_interval + ) + VALUES (?, ?, ?, ?) ON CONFLICT (destination) DO UPDATE SET + failure_ts = EXCLUDED.failure_ts, retry_last_ts = EXCLUDED.retry_last_ts, retry_interval = EXCLUDED.retry_interval WHERE @@ -212,7 +219,7 @@ class TransactionStore(SQLBaseStore): OR destinations.retry_interval < EXCLUDED.retry_interval """ - txn.execute(sql, (destination, retry_last_ts, retry_interval)) + txn.execute(sql, (destination, failure_ts, retry_last_ts, retry_interval)) return @@ -225,7 +232,7 @@ class TransactionStore(SQLBaseStore): txn, table="destinations", keyvalues={"destination": destination}, - retcols=("retry_last_ts", "retry_interval"), + retcols=("failure_ts", "retry_last_ts", "retry_interval"), allow_none=True, ) @@ -235,6 +242,7 @@ class TransactionStore(SQLBaseStore): table="destinations", values={ "destination": destination, + "failure_ts": failure_ts, "retry_last_ts": retry_last_ts, "retry_interval": retry_interval, }, @@ -245,6 +253,7 @@ class TransactionStore(SQLBaseStore): "destinations", keyvalues={"destination": destination}, updatevalues={ + "failure_ts": failure_ts, "retry_last_ts": retry_last_ts, "retry_interval": retry_interval, }, diff --git a/synapse/util/retryutils.py b/synapse/util/retryutils.py index b740913b58..a5f2fbef5c 100644 --- a/synapse/util/retryutils.py +++ b/synapse/util/retryutils.py @@ -80,11 +80,13 @@ def get_retry_limiter(destination, clock, store, ignore_backoff=False, **kwargs) # We aren't ready to retry that destination. raise """ + failure_ts = None retry_last_ts, retry_interval = (0, 0) retry_timings = yield store.get_destination_retry_timings(destination) if retry_timings: + failure_ts = retry_timings["failure_ts"] retry_last_ts, retry_interval = ( retry_timings["retry_last_ts"], retry_timings["retry_interval"], @@ -108,6 +110,7 @@ def get_retry_limiter(destination, clock, store, ignore_backoff=False, **kwargs) destination, clock, store, + failure_ts, retry_interval, backoff_on_failure=backoff_on_failure, **kwargs @@ -120,6 +123,7 @@ class RetryDestinationLimiter(object): destination, clock, store, + failure_ts, retry_interval, backoff_on_404=False, backoff_on_failure=True, @@ -133,6 +137,8 @@ class RetryDestinationLimiter(object): destination (str) clock (Clock) store (DataStore) + failure_ts (int|None): when this destination started failing (in ms since + the epoch), or zero if the last request was successful retry_interval (int): The next retry interval taken from the database in milliseconds, or zero if the last request was successful. @@ -145,6 +151,7 @@ class RetryDestinationLimiter(object): self.store = store self.destination = destination + self.failure_ts = failure_ts self.retry_interval = retry_interval self.backoff_on_404 = backoff_on_404 self.backoff_on_failure = backoff_on_failure @@ -186,6 +193,7 @@ class RetryDestinationLimiter(object): logger.debug( "Connection to %s was successful; clearing backoff", self.destination ) + self.failure_ts = None retry_last_ts = 0 self.retry_interval = 0 elif not self.backoff_on_failure: @@ -211,11 +219,17 @@ class RetryDestinationLimiter(object): ) retry_last_ts = int(self.clock.time_msec()) + if self.failure_ts is None: + self.failure_ts = retry_last_ts + @defer.inlineCallbacks def store_retry_timings(): try: yield self.store.set_destination_retry_timings( - self.destination, retry_last_ts, self.retry_interval + self.destination, + self.failure_ts, + retry_last_ts, + self.retry_interval, ) except Exception: logger.exception("Failed to store destination_retry_timings") -- cgit 1.4.1