[rabbitmq-discuss] limit number of messages buffered in memory in new persister

[alex chen]

Matthew,



> What you can do, is publish messages in parallel, using different
> channels. The server can coalesce these commits under some 

There is a requirement in our application that messages have to be published/received in order,
so we have to use single channel.

>> It seems to me the bottleneck was 
>> not fsync, because disk was not busy.
>> also i tried commenting out 
>> file:sync() from the src code and it has no effect.

> That is very curious. Also, don't do that - if it crashes at that point
> you're not going to be able to recover much data from disk ;) How big
> are the messages you're writing?

I disabled file:sync() just to test if the bottleneck is caused by fsync.
The result showed it was not.  In your previous email you mentioned using ext3 with data=ordered you could get close to 500 msg/sec.  that means it is affected by disk.  However, i still think 20/sec is slow.

thanks!

-alex