I have a (probably) fairly technical question about connection pooling with Axis2 I'd like to throw out there, in case someone else has come across this.
I'd like to deploy a POJO that connects to a DB backend through Axis2 on Tomcat, using the .aar archive to deploy. I managed to get it working with a connection pool the usual way (described on the tomcat site) by creating a resource in META_INF/context.xml and a resource-ref in the web.xml in WEB-INF.
In dev, this is not a big deal, but in production, I don't want to tinker with files that may affect other applications. For standard web development, I've been able to configure these things specifically for a webapp by editing the files in META-INF and WEB-INF within the application's directory structure. But unfortunately, I don't see a way to edit files specific to a .aar archive (in fact, I'm not sure an .aar file even has a service-specific WEB-INF or META-INF directory). This leaves me editing files the META-INF and WEB-INF for all the applications in Axis2, rather than just my own particular app.
Anyway - with that background, is there a way to do this? Or am I stuck with editing the apps in Axis2/WEB-INF and Axis2/META-INF for connection pooling?
Hope this makes sense - please ask me for any clarifications if it doesn't make sense - I'm fairly new to Axis2, still figuring it out.
First off, an .aar file does have a META-INF directory, but it's not for context.xml, but for service.xml (which does something completely different). It also doesn't have a WEB-INF/web.xml file.
But I don't really see a problem with adding appropriate entries to the respective files of the web app. You only need to do this once, and you can give the pool a distinctive name, so that there's no interference with other pools. Can you elaborate on why you don't want to go down that path?
Joined: Aug 18, 2008
<Can you elaborate on why you don't want to go down that path?>
Well, I'm certainly *willing* to go down that path, but I generally prefer to keep deployment self-contained. It's similar to why I might use the local lib for application-specific jar files, rather than putting them in the top level directory.
I know, technically, it isn't quite the same thing - because these are just elements in a config file, not different versions of a jar, so as long as I'm careful, it should be fine. But editing a config used by all apps definitely increases the chances of hosing all the apps running on that server, whereas a self-contained deployment reduces the chances of this happening. Even a minor typo can really mess everything up.
But it sounds like I'll have to do it this way if I want to use an .aar file, because there's no local WEB-INF.