[rabbitmq-discuss] Occasional slow message on a local machine

Thu Sep 6 11:29:27 BST 2012

On 09/06/2012 05:16 AM, Matthias Radestock wrote:
> Brennan,
> 
> On 06/09/12 10:11, Brennan Sellner wrote:
>> Okay, thanks.  I wasn't sure if there were any other triggers that would
>> cause Rabbit to touch the disk, such as the creation of a queue.
> 
> Creation/deletion of durable queues/exchanges/bindings *does* touch the
> disk. With an fsync to boot.

Good to know -- I suspected as much, but I wasn't certain.  The queues
and exchanges in question are definitely non-durable, though.

Other parts of the system do use durable queues and exchanges: would the
creation or binding of durable queues via a separate connection to
Rabbit have the potential to impact a connection not dealing in durable
constructs?

Also, does the fsync occur only on the initial creation of the durable
queue, exchange, or binding?  Or does Rabbit hit the disk every time a
client asserts that a durable object exists?

Thanks,

-Brennan