[Catalyst] Slowness on test server caused by reverse DNS lookups
Matt S Trout
dbix-class at trout.me.uk
Wed May 31 02:39:25 CEST 2006
Eric K. Olson wrote:
> Hi Folks!
> In writing my first Catalyst Application (and may I say THANK YOU for
> creating this framework!),
> I noticed a long delay on every request, which I traced to the reverse
> DNS lookups of the server
> and client IP addresses in Catalyst::Engine::HTTP::_socket_info. I
> patched it to have '-nodns'
> and '-noremotedns' to turn off the server and client reverse DNS
> lookups, respectively. That patch
> appears below, including a patch to the Helper that generates the
> myapp_server.pl code.
> I realize that the test server is just for getting people up and running
> quickly, but I think it
> is important that it have good response time, even in the face of badly
> configured (or impossible
> to properly configure) reverse DNS records. I initially attributed the
> delay to actual processing,
> which made Catalyst look REALLY slow. It was 100% DNS timeouts-- even
> with the test server my
> App (using TT and DBIC::Schema) is REALLY fast.
> 1. Is it appropriate to post this patch here, or should I be contacting
> the maintainer of that
> package directly?
there isn't exactly any "the maintainer"
Here is fine though, or the catalyst-dev@ list maybe be better.
> 2. I created an instance variable _options to hold the options hash,
> rather than passing it down
> into _handler() and then into _socket_data(). I was surprised it wasn't
> already an instance variable.
> Is this an OK thing to do? Is there a coding style standard somewhere?
> 3. I also created an accessor method options() in
> Catalyst::Engine::HTTP, but not a setter method. I
> was surprised that C:E:HTTP didn't inherit from a superclass providing
> generic accessor/setters for
> all the instance variables. Was this an OK thing to do?
It does. Check the ISA list again :)
> Basically, if I want my patches to be rolled into the official Catalyst,
> I need to know if I'm doing
> things appropriately. Any constructive input is welcome.
Proper accessor, make sure it applies cleanly against current subversion
trunk, and wait and see what the Engine experts think.
Plus a test case would be nice.
More information about the Catalyst