Cool I will take a look at what I did wrong.... One question though, if Websphere doesn't have a autodeploy feature for applications built as Wars, is there any hope of getting this working? Or will this solve the problem by deploying the app differently?
On Thu, 17 Mar 2005 21:35:02 +0100, Ate Douma <[EMAIL PROTECTED]> wrote: > > > Seth Ford wrote: > > I am trying it out but I think I am doing something wrong with the > > jetspeed2-layout-portlets.war.... Doesn't it still go in the WEB-INF > > deploy folder? I put it there but I am seeing > > INFO: Loading portlet application from web archive C:\apps\tomcat\5.0. > > \jetspeed\WEB-INF\deploy\jetspeed-layouts.war > > INFO: Portlet application "jetspeed": registered=true, deployed=true > > INFO: Portlet application "jetspeed" already registered. Skipping ini > > yment. > > INFO: Portlet application registration target is jetspeed ... > > INFO: Adding file:/C:/apps/tomcat/5.0.28/temp/jetspeed-jar-tmp/jetspee > > war/WEB-INF/classes/ to class path. > > INFO: Adding file:/C:/apps/tomcat/5.0.28/temp/jetspeed-jar-tmp/jetspee > > war/WEB-INF/lib/portals-bridges-velocity-0.1.jar to class path. > > INFO: Registered portlet app in the class loader registry... jetspeed > > INFO: Registered portlet app in the search engine... jetspeed > > INFO: Portlet application registration of jetspeed complete. > > INFO: Portlet app "jetspeed" successfuly (re)deployed. > > > > and I get an error coming back from the browser for the branch > Seth, > > I'm not sure which version of J2 you are testing but it certainly isn't > the deployment_refactoring branch. > Looking at the logging you provided this looks a cvs head version to me. > The deployment_refactoring doesn't need temporary deployment folders like > /temp/jetspeed-jar-tmp anymore and furthermore the jetspeed-layouts local > pa is registered under its own name: jetspeed-layouts (which means I had > to change the psml definitions for that too). > > > > > Encountered the following problem(s) while attmepting to render > > portlet fragment: dp-1 > > Failed to retrieve Portlet Definition for jetspeed::VelocityTwoColumns > > org.apache.jetspeed.container.window.FailedToRetrievePortletWindow: No > > PortletEntity exists for for id dp-1 removing window from cache. > > Failed to retrieve Portlet Definition for jetspeed::VelocityTwoColumns > > org.apache.jetspeed.container.window.FailedToRetrievePortletWindow: No > > PortletEntity exists for for id dp-1 removing window from cache. > > > > > > > > On Thu, 17 Mar 2005 03:01:59 +0100, Ate Douma <[EMAIL PROTECTED]> wrote: > > > >>Dear all, > >> > >>Today I committed a big update for the deployment refactoring branch. > >>I've added the following features: > >> > >>- Moved Deployment interfaces and related components to the jetspeed-api > >>subproject, > >> as well as the ApplicationServerManager interface and its new Result > >> component. > >> This allows access to these services for portlet applications. > >> > >>- Provided a new ManagerServlet somewhat like the ManagerServlet of Tomcat. > >> It allows remote control of portlet applications and the registry with the > >> following functions: start, stop, reload, list, undeploy and deploy > >> (upload). > >> I also created a new JetspeedConsole CLI using the ManagerServlet which > >> works > >> well, but which I haven't committed yet because I need to clean it up > >> first. > >> > >>- The ManagerServlet depends for several of its tasks on a > >>ApplicationServerManager > >> implementation. With the TomcatManager all of its features can now be > >> used. > >> The implementations of the JBossManager and WeblogicManager are still > >> empty > >> shells though and probably someone else with more knowledge of these > >> application servers should take a look at those. > >> > >>- A new PortletApplicationManager portlet. Yes, another PAM indeed, and the > >> naming of these is getting confusing. > >> This new portlet though provides (almost) the same functionality as the > >> ManagerServlet. > >> Its lists all registered portlet applications in a table with action > >> links for: > >> start, stop, undeploy and delete (unregister from the registry). > >> It also shows if an portlet application actually is running or not. > >> Because I wanted to provide this functionality in table layout, I decided > >> to not > >> implement it in the already existing PAM (PortletApplicationBrowser) > >> portlet which > >> presents the portlet applications and its portlets in a tree. > >> For now, this portlet is accessible from the Administrative folder > >> (pam2.psml). > >> We probably need to discuss though if and/or how these two portlets > >> should be > >> integrated. > >> This portlet also depends on a ApplicationServerManager. If none is > >> configured (like > >> for JBoss) it will still show if a portlet application is running or not > >> and allow > >> to delete (unregister) an portlet application. > >> > >>- J2 on JBoss > >> Because one of the premises of this deployment refactoring was that it > >> should make > >> it easier to deploy J2 on other application servers as well, I decided to > >> prove this > >> and got my feet wet trying it out for JBoss (3.2.7). > >> > >> With success! > >> I wrote a list of things to do to get this branch running on JBoss in a > >> comment to > >> JS2-210: http://issues.apache.org/jira/browse/JS2-210#action_60983 > >> Once this branch is merged into the head branch I'll provide proper > >> instructions on > >> the website or wiki (although I'm not much of a wiki jockey yet). > >> I also tested with JBoss 4.0.1sp1 which looks to be working just as well. > >> I didn't > >> have time to test that one a lot though. > >> > >> Of course, there were quite a few problems to solve, but *none* were > >> related to > >> (re)deployment. And many of the problems as indicated by the wiki pages > >> for JBoss > >> deployment of the M1 or current head version of J2 seems to be resolved > >> :-) > >> > >> The following issues are important though: > >> - commons-logging and Log4J dependencies > >> JBoss provides commons-logging and Log4J, as well as the Log4J > >> configuration from > >> a shared classloader to all the applications. > >> This really is conflicting with the way we use them in Jetspeed-2 under > >> Tomcat > >> (although Tomcat 5.5 also is giving more headaches with this now). > >> One simply cannot have the commons-logging and log4j jars anymore in > >> WEB-INF/lib > >> because it definitely is giving classloader problems under JBoss. > >> > >> Furthermore, our own log4j configuration conflicts with the global > >> configuration > >> provided with JBoss. If you initialize a new log4j configuration from a > >> web application > >> like we do with J2, you end up closing and detaching existing loggers > >> and appenders > >> and rerouting them into the J2 logging. Likewise, if some other web > >> application > >> deployed after J2 does the same, the J2 logging might get closed and > >> detached and > >> rerouted into this other web application its logging. > >> > >> Anyway you look at it, dynamic log4j configuration is very problematic > >> under JBoss. > >> And, while scanning the internet for a way around this, I found out > >> many others > >> encountered the same problem and not only on JBoss but other > >> application servers > >> as well. > >> > >> After a long investigation of the concrete implementations of > >> commons-logging and > >> log4j though I've come up with a solution which I called > >> IsolatedLog4JLogger. > >> As I put a lot of javadoc into that class I'm not going to reproduce it > >> here. > >> But, I invite everyone to have a look at it. Its currently only > >> committed to the > >> deployment_refactoring branch under > >> portal/src/java/org/apache/jetspeed/util. > >> > >> Whats left is a future solution of the packaging of these jars. Having > >> to remove > >> them after a war is build is quite dumb and easy to forget as well. > >> > >> - JBoss UniversalClassLoader > >> I don't think I have to explain anyone interested in JBoss what this > >> means. > >> Anyway, with a few tweaks to the JBoss configurations I managed to get > >> it working > >> "properly" ;-) Other solutions are also possible (like providing a > >> jboss-web.xml > >> in *all* portlet application war files). > >> > >> - JAAS authentication > >> This was quite simple to implement but in the end turned out to be > >> working only > >> half way. > >> The authentication itself is easy to do by configuring our > >> DefaultLoginModule in > >> the JBoss login-config.xml. > >> But, the propagation of the Subject principals to JBoss is quite a > >> problem. > >> Simply put: none is. > >> JBoss requires the Subject to be configured with a > >> java.security.acl.Group named > >> "Roles" containing all the "Role" principals. As our security > >> principals don't > >> implement java.securitry.acl.Group (yet) none is recognized by JBoss. > >> After some further investigation, it turns out other application servers > >> might require a Group implementation as well. So, we might need to > >> refactor our > >> security principals a bit to get this to work. > >> As we don't actually *use* request.isUserInRole(roleName) in our demo > >> portlets > >> yet (except for the security demo portlet), the default J2 deployment > >> actually > >> works quite good. > >> > >>Well, I'd be very interested in others experiences with this branch (the > >>latest and > >>from before my last changes). I know some are trying it out on different > >>platforms as > >>well. > >> > >>Seth Ford is still unsuccessful with getting M1 running on WebSpere. Seth, > >>I can't > >>really help you out with WebSphere but I really would suggest you try it > >>out with the > >>latest from this branch (check the JBoss deployment instructions as > >>reference). > >> > >>I think that with a few success stories of other deployments on different > >>platforms > >>of this branch it might be almost ready for merging with the head. > >>Deleting (unregistering) of portlet applications is now provided for, as > >>well as > >>start, stop, (remote) deployment and undeployment for Tomcat. > >> > >>If anyone has some requirements which should be fixed/implemented on this > >>branch > >>before considering it ready for merge with the head (besides more proof it > >>works > >>too on other platforms), please let me know. If possible and reasonable, > >>I'll try > >>to provide for it. > >> > >>Regards, Ate > >> > >>--------------------------------------------------------------------- > >>To unsubscribe, e-mail: [EMAIL PROTECTED] > >>For additional commands, e-mail: [EMAIL PROTECTED] > >> > >> > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
