diff options
| author | Matthew Sackman <matthew@lshift.net> | 2009-08-04 12:27:50 +0100 |
|---|---|---|
| committer | Matthew Sackman <matthew@lshift.net> | 2009-08-04 12:27:50 +0100 |
| commit | da2f2f438f55f969e8c0b2af466750a017a578e2 (patch) | |
| tree | 1e5591165830e946930179bd0fe729b4958b877c | |
| parent | 8a3dfd9ccb8d7b097804e4f960ad3237d2cc213f (diff) | |
| download | rabbitmq-server-git-da2f2f438f55f969e8c0b2af466750a017a578e2.tar.gz | |
Done. Testing the latency between the call to conserve_memory and when the msg gets processed shows this is benficial:
In the erlang client, hammer in a few million messages, with no consumer. This causes the channel mailbox to get pretty big. Without the higher priority can demonstrate a delay of over a second before the conserve_message gets processed. With this, it's a mere fraction of that.
| -rw-r--r-- | src/rabbit_channel.erl | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/rabbit_channel.erl b/src/rabbit_channel.erl index 3089bb6293..87664de30b 100644 --- a/src/rabbit_channel.erl +++ b/src/rabbit_channel.erl @@ -89,7 +89,7 @@ deliver(Pid, ConsumerTag, AckRequired, Msg) -> gen_server2:cast(Pid, {deliver, ConsumerTag, AckRequired, Msg}). conserve_memory(Pid, Conserve) -> - gen_server2:cast(Pid, {conserve_memory, Conserve}). + gen_server2:pcast(Pid, 9, {conserve_memory, Conserve}). %%--------------------------------------------------------------------------- |