[Rabbit-dev] Logging Usage

[Robert Olofsson]

Hi!

On Fri, 19 Aug 2011 01:44:56 -0400
Luis Soltero <lsoltero at globalmarinenet.com> wrote:
> working!

Great!

> Can you tell me more about network_write/read?

I can try.

You have 4 different directions for data in rabbit: 
client, network, cache and proxy.

The client is the traffic to/from the client web browser.
Network is for traffic between rabbit and upstream server.
Cache is for traffic between rabbit and the cache.
Proxy is a special case for resources generated by rabbit, like
the rabbit status pages.

All of these 4 traffic loggers have 4 values:
read, write, transfer_to, transfer_from.

Read and write are normal read/write.
The transfer values are zero copy data transfers.

For a normal non cached request you will see:
client read and client write as well as network read and write.
If the cache is allowed to cache the resource you will also see
cache writes.

For a cached resource you will see:
client read client write for the header handling and 
client transfer to + cache transfer from (assuming your cache supports
transfer, but it ought to).

Things get more interesting if you hit a cached resource,
but with a forced cache validation. Then you will see
client read, client write, network read, network write and
cache to client transfer.

Does this make sense?

/robo