track replication destination health, and perform exponential back-off when sending transactions. does *not* yet retry transactions, but drops them on the floor if waiting for a server to recover.

1 files changed, 36 insertions, 7 deletions

diff --git a/synapse/federation/replication.py b/synapse/federation/replication.py

index 01f87fe423..f9c05b5ea3 100644
--- a/synapse/federation/replication.py
+++ b/synapse/federation/replication.py
@@ -723,6 +723,8 @@ class _TransactionQueue(object):
         deferreds = []
 
         for destination in destinations:
+            # XXX: why don't we specify an errback for this deferred
+            # like we do for EDUs? --matthew
             deferred = defer.Deferred()
             self.pending_pdus_by_dest.setdefault(destination, []).append(
                 (pdu, deferred, order)
@@ -738,6 +740,9 @@ class _TransactionQueue(object):
     # NO inlineCallbacks
     def enqueue_edu(self, edu):
         destination = edu.destination
+        
+        if destination == self.server_name:
+            return
 
         deferred = defer.Deferred()
         self.pending_edus_by_dest.setdefault(destination, []).append(
@@ -766,14 +771,23 @@ class _TransactionQueue(object):
         )
 
         yield deferred
-
+        
     @defer.inlineCallbacks
     @log_function
     def _attempt_new_transaction(self, destination):
+        
+        (retry_last_ts, retry_interval) = self.store.get_destination_retry_timings(destination)
+        if retry_last_ts + retry_interval > int(self._clock.time_msec()):
+            logger.info("TX [%s] not ready for retry yet - dropping transaction for now")
+            return
+        
         if destination in self.pending_transactions:
+            # XXX: pending_transactions can get stuck on by a never-ending request
+            # at which point pending_pdus_by_dest just keeps growing.
+            # we need application-layer timeouts of some flavour of these requests
             return
 
-        #  list of (pending_pdu, deferred, order)
+        # list of (pending_pdu, deferred, order)
         pending_pdus = self.pending_pdus_by_dest.pop(destination, [])
         pending_edus = self.pending_edus_by_dest.pop(destination, [])
         pending_failures = self.pending_failures_by_dest.pop(destination, [])
@@ -781,7 +795,8 @@ class _TransactionQueue(object):
         if not pending_pdus and not pending_edus and not pending_failures:
             return
 
-        logger.debug("TX [%s] Attempting new transaction", destination)
+        logger.debug("TX [%s] Attempting new transaction (pdus: %d, edus: %d, failures: %d)",
+            destination, len(pending_pdus), len(pending_edus), len(pending_failures))
 
         # Sort based on the order field
         pending_pdus.sort(key=lambda t: t[2])
@@ -814,7 +829,7 @@ class _TransactionQueue(object):
             yield self.transaction_actions.prepare_to_send(transaction)
 
             logger.debug("TX [%s] Persisted transaction", destination)
-            logger.debug("TX [%s] Sending transaction...", destination)
+            logger.info("TX [%s] Sending transaction [%s]", destination, transaction.transaction_id)
 
             # Actually send the transaction
 
@@ -835,6 +850,8 @@ class _TransactionQueue(object):
                 transaction, json_data_cb
             )
 
+            logger.info("TX [%s] got %d response", destination, code)
+
             logger.debug("TX [%s] Sent transaction", destination)
             logger.debug("TX [%s] Marking as delivered...", destination)
 
@@ -849,6 +866,7 @@ class _TransactionQueue(object):
                 if code == 200:
                     deferred.callback(None)
                 else:
+                    start_retrying(destination, retry_interval)
                     deferred.errback(RuntimeError("Got status %d" % code))
 
                 # Ensures we don't continue until all callbacks on that
@@ -861,12 +879,12 @@ class _TransactionQueue(object):
             logger.debug("TX [%s] Yielded to callbacks", destination)
 
         except Exception as e:
-            logger.error("TX Problem in _attempt_transaction")
-
             # We capture this here as there as nothing actually listens
             # for this finishing functions deferred.
-            logger.exception(e)
+            logger.exception("TX [%s] Problem in _attempt_transaction: %s", destination, e)
 
+            start_retrying(destination, retry_interval)
+            
             for deferred in deferreds:
                 if not deferred.called:
                     deferred.errback(e)
@@ -877,3 +895,14 @@ class _TransactionQueue(object):
 
             # Check to see if there is anything else to send.
             self._attempt_new_transaction(destination)
+
+def start_retrying(destination, retry_interval):
+    # track that this destination is having problems and we should
+    # give it a chance to recover before trying it again
+    if retry_interval:
+        retry_interval *= 2
+    else:
+        retry_interval = 2 # try again at first after 2 seconds
+    self.store.set_destination_retry_timings(destination,
+        int(self._clock.time_msec()), retry_interval)
+    
\ No newline at end of file