Thank you for the info. I am afraid your reply has generated more questions :)
Where does the connect_on_init call and the definition of $_db_config go? In startup.pl? Or in the handler module? I don't think I really understand how the whole system fits together. Is the startup.pl code run once then shared by all threads? What about code that is in the handler module but not in the handler subroutine? Eg the module loads etc. Are they called once per thread up to maxthreads? Or just once per module load and shared between the threads. Thanks for your help, much appreciated Matt Sent from my GoodLink synchronized handheld (www.good.com) -----Original Message----- From: Jonathan Vanasco [mailto:[EMAIL PROTECTED] Sent: Friday, June 01, 2007 05:31 PM Pacific Standard Time To: Matt Williamson Cc: modperl@perl.apache.org Subject: Re: Dbi best practice question On Jun 1, 2007, at 7:47 PM, Matt Williamson wrote: > Hi > > When using a MySQL database with modperl 2. I can find lots of docs on > using DBI, but not much on what is the best practice... > > I am currently connecting, using the db and exiting on each > request, and > relying on Apache::DBI to cache the connection. E.g. well, the first thing i would change, is to add this: Apache::DBI->connect_on_init( $_db_config{'dbi'}, $_db_config{'dbUser'}, $_db_config{'dbPass'}, $_db_config{'dbArgs'} ); that''ll instantiate a connection after the fork -- so you don't have the connection overhead on the first connect. you still keep the DBI->connect -- this just preloads it. you also might want to add a cleanup handler that just issues a rollback, which is something that perrin often suggests, and I admittedly blindly follow. i trust him on those things though :) > If I understand Apache::DBI correctly, the DBI->connect picks up a > cached connection. Is it possible to prepare queries in one place and > reuse in all the child nodes? E.g. using prepare_cached. If so > where do > I put that? In the startup file? In the same module? i have no idea how to handle the prepared queries, but i will say this; i'm pretty sure that in mysql prepared queries work per connection ( i know they do in pgsql ) -- so you wouldn't be able to do it in startup i blieve db connections close/respawn when a child dies / is born ( ie maxrequests ) so you'll probably have to code for something to handle populating the statements. // Jonathan Vanasco | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - | SyndiClick.com | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - | FindMeOn.com - The cure for Multiple Web Personality Disorder | Web Identity Management and 3D Social Networking | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - | RoadSound.com - Tools For Bands, Stuff For Fans | Collaborative Online Management And Syndication Tools | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
