Return-Path: Delivered-To: apmail-incubator-geronimo-dev-archive@www.apache.org Received: (qmail 78348 invoked from network); 3 Oct 2003 23:27:43 -0000 Received: from daedalus.apache.org (HELO mail.apache.org) (208.185.179.12) by minotaur-2.apache.org with SMTP; 3 Oct 2003 23:27:43 -0000 Received: (qmail 85718 invoked by uid 500); 3 Oct 2003 23:27:21 -0000 Delivered-To: apmail-incubator-geronimo-dev-archive@incubator.apache.org Received: (qmail 85668 invoked by uid 500); 3 Oct 2003 23:27:21 -0000 Mailing-List: contact geronimo-dev-help@incubator.apache.org; run by ezmlm Precedence: bulk list-help: list-unsubscribe: list-post: Reply-To: geronimo-dev@incubator.apache.org Delivered-To: mailing list geronimo-dev@incubator.apache.org Received: (qmail 85644 invoked from network); 3 Oct 2003 23:27:21 -0000 Received: from unknown (HELO hotmail.com) (207.68.163.101) by daedalus.apache.org with SMTP; 3 Oct 2003 23:27:21 -0000 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Fri, 3 Oct 2003 16:27:28 -0700 Received: from 144.139.159.171 by sea1fd.sea1.hotmail.msn.com with HTTP; Fri, 03 Oct 2003 23:27:28 GMT X-Originating-IP: [144.139.159.171] X-Originating-Email: [gianny_damour@hotmail.com] From: "gianny DAMOUR" To: geronimo-dev@incubator.apache.org Bcc: Subject: Re: [Deployment] IM Summary for Directory Issue Date: Sat, 04 Oct 2003 01:27:28 +0200 Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed Message-ID: X-OriginalArrivalTime: 03 Oct 2003 23:27:28.0688 (UTC) FILETIME=[E91BAB00:01C38A05] X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N X-Spam-Rating: minotaur-2.apache.org 1.6.2 0/1000/N On Fri, 3 Oct 2003 17:32:28 -0400 (EDT), Aaron Mulder wrote: >1) We will save all distributed files to one directory, and we will set >that to be the same as the "deploy" directory by default (keeping all >deployments in the same location, whether by user-copied files or by >JSR-88). DeploymentScanner implements a JMX managed operation in order to add a URL to its scanner. So, I've got the feeling that one can distribute where we want. >2) For each application module in the deploy directory, we will save the >Geronimo-specific DD outside the module, with a name based on the module >name ("foo.war" gets the DD "geronimo-deployment-foo.war.xml" or something >along those lines). I agree on the general idea. However, what do you think of the following? It seems to me that we will need rather quickly a mean to persist J2EEManagedObject. Based on previous mails, Dain has this task on its to-do and will support. Indeed, if I understand correctly the problem, one only needs to persist the state of various components between server start-ups. For instance, the DeploymentManager distributes an application component to a specific server and puts it in a "distribution" directory. When one requests the start of this specific component, one could simply add to the DeploymentScanner a new URL referencing this "distribution" directory. The DeploymentScanner will then do its job as expected. Now the problem is, what happen when the server is shut-down? Does it mean that I will have to reconfigure the DeploymentScanner? I think that a possible answer is to persist the set of URL scanned by the DeploymentManager and re-use this persisted state to re-configure the DeploymentScanner on server start-up. This problem is the same one for all the application components and can be answered by persisting the J2EEManagedObject abstracting them. For instance, when a RAR is started, one can call a "Persistence Service" on the MBean mirroring this RAR and this service will persist somewhere and somehow that the new state of this ManagedObject. I really like the idea of putting this state in an easy to read XML file (why not an enhanced DD). However, one could also persist via standard serialization the state but it means that we will need to support a tool to edit easily such serialized instances. However, what do you think about defining a single XML repository? I see some advantages: For instance, to implement the DeploymentManager.getAvailableModules is trivial: one reads the repository and that is it. If one have multiple repository (multiple XML files), a possible implementation is to query directly the MBeanServer and even if it is easy to implement, an end-user will not be able to search via a standard text editor what is by now deployed. In other words, I agree on the fact that one will need to persist the state of server components between start-up. However, I think that this is the responsibility of a "Persistence Service" to tackle this problem. Gianny _________________________________________________________________ MSN Messenger 6 http://g.msn.fr/FR1001/866 : ajoutez une image � votre pseudo pour dialoguer avec vos amis