Return-Path: X-Original-To: apmail-geronimo-dev-archive@www.apache.org Delivered-To: apmail-geronimo-dev-archive@www.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 69039BF99 for ; Wed, 4 Jan 2012 09:40:41 +0000 (UTC) Received: (qmail 43124 invoked by uid 500); 4 Jan 2012 09:40:39 -0000 Delivered-To: apmail-geronimo-dev-archive@geronimo.apache.org Received: (qmail 42707 invoked by uid 500); 4 Jan 2012 09:40:19 -0000 Mailing-List: contact dev-help@geronimo.apache.org; run by ezmlm Precedence: bulk list-help: list-unsubscribe: List-Post: Reply-To: dev@geronimo.apache.org List-Id: Delivered-To: mailing list dev@geronimo.apache.org Received: (qmail 42698 invoked by uid 99); 4 Jan 2012 09:40:15 -0000 Received: from nike.apache.org (HELO nike.apache.org) (192.87.106.230) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 04 Jan 2012 09:40:15 +0000 X-ASF-Spam-Status: No, hits=2.8 required=5.0 tests=HTML_MESSAGE,RCVD_IN_DNSWL_LOW,SPF_PASS,URI_HEX X-Spam-Check-By: apache.org Received-SPF: pass (nike.apache.org: domain of forrestxm@gmail.com designates 209.85.214.182 as permitted sender) Received: from [209.85.214.182] (HELO mail-tul01m020-f182.google.com) (209.85.214.182) by apache.org (qpsmtpd/0.29) with ESMTP; Wed, 04 Jan 2012 09:40:07 +0000 Received: by obbwd18 with SMTP id wd18so19976472obb.13 for ; Wed, 04 Jan 2012 01:39:46 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; bh=AJ/JKbtQkWrMkU5+9LOP+CmEi9FTefFZ8mnZEP2q+/8=; b=qweZ16PEAZpp9pswA+VEjHHw+3RnnNQsrbfSIBU7s6E8hOMmFaxtKYGdhtTphE+oFq RxapTWI1+ZD0v83V+TWFpt2fqdgEduM5eoljCtpnpug1Pi5DIAGkJMDIl8mRwtjZ9llr uEv0NHhCE9tm/m0ZEF9+unBQPbWz/AUWF+UyM= MIME-Version: 1.0 Received: by 10.182.147.106 with SMTP id tj10mr9608990obb.12.1325669986763; Wed, 04 Jan 2012 01:39:46 -0800 (PST) Received: by 10.182.47.10 with HTTP; Wed, 4 Jan 2012 01:39:46 -0800 (PST) In-Reply-To: References: <4A5E63D9-87FB-4408-8BA3-440D757A3A70@yahoo.com> <74DEC439-60C9-488B-9A96-71AFBA25F734@yahoo.com> <00692F05-4BEA-4272-B11B-A04969583159@yahoo.com> Date: Wed, 4 Jan 2012 17:39:46 +0800 Message-ID: Subject: Re: Geronimo 3 and karaf 3 From: Forrest Xia To: dev@geronimo.apache.org Content-Type: multipart/alternative; boundary=f46d04446c57a6837504b5b0981b X-Virus-Checked: Checked by ClamAV on apache.org --f46d04446c57a6837504b5b0981b Content-Type: text/plain; charset=ISO-8859-1 On Wed, Jan 4, 2012 at 5:32 PM, Forrest Xia wrote: > I got some time to try the trunk change locally after G 2.1.8 release > stuffs are basically done :-) > > Several findings about the trunk changes: > 1. The server trunk can only be compiled with Maven 3, that's the reason > it failed to compile in the TCK environment(the maven version used in TCK > env is 2.2.1). I will update the TCK env in the coming days. > 2. The testsuite cannot be executed since geronimo-maven-plugin cannot > launch the server now. This issue will also block TCK execution. > 3. The command line interfaces has big changes, I cannot find > startup/shutdown/geronimo scripts now. I do not look into the changes > further, so I even don't know how to launch the server now. > Sorry, I reread the mail thread and got known that how to start the server now: bin/karaf -l > > So generally, I think we need a introduction about how to use the trunk > build now. > > > On Thu, Dec 22, 2011 at 5:14 PM, Forrest Xia wrote: > >> >> >> On Thu, Dec 22, 2011 at 2:27 AM, David Jencks wrote: >> >>> I pushed a karaf snapshot.... maybe that will help? I don't see this >>> problem locally. >>> >> No help for server build in AHP, I will try it locally. thanks! >> >> >>> >>> thanks >>> david jencks >>> >>> On Dec 21, 2011, at 8:18 AM, Forrest Xia wrote: >>> >>> Trunk build failed with this error when building a new module *Geronimo >>> Framework, Feature :: DS and Metatype* >>> >>> [INFO] Internal error in the plugin manager executing goal >>> 'org.apache.karaf.tooling:karaf-maven-plugin:3.0.0-SNAPSHOT:features-generate-descriptor': >>> Unable to find the mojo 'features-generate-descriptor' (or one of its >>> required components) in the plugin >>> 'org.apache.karaf.tooling:karaf-maven-plugin' >>> Component descriptor cannot be found in the component repository: >>> org.sonatype.aether.RepositorySystem. >>> >>> >>> On Tue, Dec 20, 2011 at 2:06 PM, David Jencks wrote: >>> >>>> BTW, to get the regions/isolation stuff working I think we are going to >>>> need to replace our use of BundleListener/SynchronousBundleListener with >>>> the (updated-for-4.3) aries RecursiveBundleTracker. I think we'll need >>>> also change from ConfigurationActivator to an extender pattern. I'd guess >>>> the ConfigurationActivator functionality could be moved to >>>> DependencyManager rather than having an additional tracker. >>>> >>>> thanks >>>> david jencks >>>> >>>> On Dec 20, 2011, at 10:13 AM, David Jencks wrote: >>>> >>>> OK, I just committed this stuff, with reference to GERONIMO-6240 >>>> . >>>> >>>> Some more hints.... >>>> >>>> I can build all the way through with >>>> MAVEN_OPTS="-XX:MaxPermSize=2048m -Xms2048m -Xmx4096m" >>>> >>>> I can start karaf after setting >>>> >>>> export JAVA_MAX_MEM=2048m >>>> export JAVA_MAX_PERM_MEM=512m >>>> >>>> The car packaging is set up to stop and wait if it gets stuck. In an >>>> earlier version of this you'd get the karaf console and you could use karaf >>>> commands to investigate what was going on. For some reason this isn't >>>> working now. If you get into this situation, you need to kill the maven >>>> java process some way. Usually setting a breakpoint at DependencyManager >>>> line 571 will show you a bundle that has a resolution problem that you can >>>> then fix. >>>> >>>> The problem with the console deploy-type commands I think relates to >>>> using the karaf RMIRegistry. I'm going to modify it so it includes the >>>> port as a service property, then we can look for the osgi service and get >>>> its port instead of the port gbean attribute. >>>> >>>> thanks >>>> david jencks >>>> >>>> On Dec 19, 2011, at 9:10 PM, David Jencks wrote: >>>> >>>> more not-yet-working inline >>>> >>>> On Dec 19, 2011, at 5:08 PM, David Jencks wrote: >>>> >>>> I've been spending a lot of time working to rebase geronimo on karaf 3 >>>> so we can have a maintainable future and get stuff like osgi 4.3, up to >>>> date aries components, and the experimental region support now in karaf. >>>> >>>> >>>> After a lot of work I have everything except clustering building and >>>> after turning off a couple problematic modules the tomcat-javaee6 server >>>> starts and the web admin console appears to work at least a little bit. >>>> I'd like a little vacation this year and would like to commit this work >>>> first so that others can help with the loose ends if they like. I'll >>>> probably be around to answer questions in any case. >>>> >>>> >>>> The modules that don't start are: >>>> >>>> >>>> activemq-broker-blueprint. The ${X + Y} stuff is not getting evaluated >>>> at all. I don't know if this is an xbean-blueprint problem or an aries >>>> blueprint problem or a side effect of running in geronimo. >>>> >>>> As a result activemq-ra and tomcat-console-activemq can't be started. >>>> >>>> >>>> client-deployer. I think this is a pretty simple gbean name problem >>>> but I haven't looked into it. >>>> >>>> >>>> >>>> Here are some of the changes: >>>> >>>> >>>> -- assemble the server using a combination of karaf assembly from >>>> features and kars and geronimo assembly from geronimo plugins. We now use >>>> the same base karaf assembly stuff as the normal default full karaf >>>> assembly (except I might have left out the spring feature repository). >>>> >>>> >>>> -- basic geronimo components such as the kernel, configuration manager, >>>> dependency manager, deployer, and service config builder are set up as osgi >>>> declarative services so they start without any geronimo configuration. >>>> They are generally configured through config admin as appropriate. Most >>>> of these also have gbean wrappers so they can be accessed through gbean >>>> references. >>>> >>>> >>>> -- "geronimo" is started from a DS component, EmbeddedDaemon. >>>> >>>> >>>> -- I think I'm using the karaf remote jmx security rather than ours. >>>> The capabilities are similar but not identical. >>>> >>>> >>>> Some other things that are not working yet: >>>> >>>> >>>> -- The (gogo) geronimo console commands that work through "remote" >>>> gbean proxies don't work AFAIK. Probably one way to fix this would be to >>>> expose some more of the DS components using gbean wrappers, but I haven't >>>> looked into this yet. >>>> >>>> >>>> -- the app client (as well as the client-deployer) is not working yet >>>> at all. We may be able to use command line args to tell the EmbeddedDaemon >>>> it's an app client, or possibly not. We may be able to use a karaf >>>> instance to supply different ConfigAdmin settings to e.g. the local >>>> attribute manager to convince it it's an app client. Similarly the >>>> separate console-like things presumably won't work either. >>>> >>>> >>>> -- the EditableConfigurationManager needs to be replaced by a separate >>>> component that edits the configuration it gets from the normal >>>> configuration manager. I think this affects some part of the admin console. >>>> >>>> >>>> --I couldn't get the xml stream 1.2 and jaxb 2.2 to work with the spec >>>> jars as bundles. According to >>>> http://servicemix.396122.n5.nabble.com/DISCUSS-Enhance-specs-to-work-better-with-JRE-td5001108.htmleven if you do get them to work (as we seem to have up to now by not >>>> exposing the packages from the framework) that breaks other stuff. I think >>>> we need to investigate the karaf-activator stuff guillaume wrote and adapt >>>> our specs to use it. At the moment I have the framework lying and claiming >>>> later versions for the xmlstream and jaxb packages. I haven't found any >>>> documentation for karaf-activator yet. >>>> >>>> -- the build uses a lot more memory. I typically run out of permgen >>>> twice during the build with MAVEN_OPTS = -XX:MaxPermSize=512m -Xms1024m >>>> -Xmx2048m >>>> >>>> -- startup AFAIK only works as ./bin/karaf -l rather than our geronimo >>>> scripts. Again, I have to increase memory settings for the server to fully >>>> start. >>>> >>>> >>>> >>>> I'f there's no strong opposition I'd like to commit this tomorrow. >>>> >>>> >>>> Many thanks >>>> >>>> david jencks >>>> >>>> >>>> >>>> david jencks >>>> >>>> >>>> >>>> >>> >>> >>> -- >>> Thanks! >>> >>> Regards, Forrest >>> >>> >>> >> >> >> -- >> Thanks! >> >> Regards, Forrest >> >> > > > -- > Thanks! > > Regards, Forrest > > -- Thanks! Regards, Forrest --f46d04446c57a6837504b5b0981b Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable

On Wed, Jan 4, 2012 at 5:32 PM, Forrest = Xia <forrestxm@= gmail.com> wrote:
I got some time to try the trunk change locally after G 2.1.8 release stuff= s are basically done :-)

Several findings about the trunk changes:1. The server trunk can only be compiled with Maven 3, that's the rea= son it failed to compile in the TCK environment(the maven version used in T= CK env is 2.2.1). I will update the TCK env in the coming days.
2. The testsuite cannot be executed since geronimo-maven-plugin cannot laun= ch the server now. This issue will also block TCK execution.
3. The comm= and line interfaces has big changes, I cannot find startup/shutdown/geronim= o scripts now. I do not look into the changes further, so I even don't = know how to launch the server now.
Sorry, I reread the mail thread and got known that how to= start the server now: bin/karaf -l

So generally, I think we need a introduction about how to use the trunk= build now.


On Thu, Dec 22, 2011 at 5:14 PM, Forrest Xia &= lt;forrestxm@gmail= .com> wrote:


On Thu, Dec 22, 2011 at 2:27 AM, David Jencks <david_j= encks@yahoo.com> wrote:
I pushed a karaf snapshot.... maybe tha= t will help? =A0I don't see this problem locally.
No help for server build in AHP, I will try it locally. thanks!
=A0

thanks
david jencks

On Dec 21, 201= 1, at 8:18 AM, Forrest Xia wrote:

Trunk = build failed with this error when building a new module Geronimo Framewo= rk, Feature :: DS and Metatype

=A0[INFO] Internal=20 error in the plugin manager executing goal=20 'org.apache.karaf.tooling:karaf-maven-plugin:3.0.0-SNAPSHOT:features-ge= nerate-descriptor': Unable to find the mojo 'features-generate-descriptor' (or one of = its=20 required components) in the plugin=20 'org.apache.karaf.tooling:karaf-maven-plugin'
=A0Componen= t descriptor cannot be found in the component repository: org.sonatype.aeth= er.RepositorySystem.


On Tue, Dec 20, 2011 at 2:06 PM, David Jencks <david_jencks@yahoo.com= > wrote:
BTW, to get the regions/isolation stuff= working I think we are going to need to replace our use of BundleListener/= SynchronousBundleListener with the (updated-for-4.3) aries RecursiveBundleT= racker. =A0I think we'll need also change from ConfigurationActivator t= o an extender pattern. =A0I'd guess the ConfigurationActivator function= ality could be moved to DependencyManager rather than having an additional = tracker.

thanks
david jencks<= /div>

On Dec 20, 2011, at 10:13 A= M, David Jencks wrote:

OK, I just committed this stuff, with r= eference to=A0GERONIMO-6240.

Some more hints....

I can build all the way through with=A0
= MAVEN_OPTS=3D"-XX:MaxPermSize=3D2048m -Xms2048m -Xmx4096m"
<= /div>

I can start karaf after setting

export JAVA_MAX_MEM=3D2048m
export JAVA_MAX_PERM_MEM=3D512m<= /div>

The car packaging is set up to stop and wait if it= gets stuck. =A0In an earlier version of this you'd get the karaf conso= le and you could use karaf commands to investigate what was going on. =A0Fo= r some reason this isn't working now. =A0If you get into this situation= , you need to kill the maven java process some way. =A0Usually setting a br= eakpoint at DependencyManager line 571 will show you a bundle that has a re= solution problem that you can then fix.

The problem with the console deploy-type commands I thi= nk relates to using the karaf RMIRegistry. =A0I'm going to modify it so= it includes the port as a service property, then we can look for the osgi = service and get its port instead of the port gbean attribute.

thanks
david jencks

O= n Dec 19, 2011, at 9:10 PM, David Jencks wrote:

more not-yet-working inline

On Dec 19, 2011, at 5:08 = PM, David Jencks wrote:

I've been spending a lot of time working = to rebase geronimo on karaf 3 so we can have a maintainable future and get = stuff like osgi 4.3, up to date aries components, and the experimental regi= on support now in karaf.

After a lot of work I have everything except clustering building and= after turning off a couple problematic modules the tomcat-javaee6 server s= tarts and the web admin console appears to work at least a little bit. =A0= =A0I'd like a little vacation this year and would like to commit this w= ork first so that others can help with the loose ends if they like. =A0I= 9;ll probably be around to answer questions in any case.

The modules that don't start are:

activemq-broker-blue= print. =A0The ${X + Y} stuff is not getting evaluated at all. =A0I don'= t know if this is an xbean-blueprint problem or an aries blueprint problem = or a side effect of running in geronimo.
As a result activemq-ra and tomcat-c= onsole-activemq can't be started.

client-deployer. =A0I think= this is a pretty simple gbean name problem but I haven't looked into i= t.


Here are some of the chan= ges:

-- assemble the server using a combination of karaf assembly from features = and kars and geronimo assembly from geronimo plugins. =A0We now use the sam= e base karaf assembly stuff as the normal default full karaf assembly (exce= pt I might have left out the spring feature repository).

-- basic geronimo components such as the kernel, configuration manag= er, dependency manager, deployer, and service config builder are set up as = osgi declarative services so they start without any geronimo configuration.= =A0They are generally configured through config admin as appropriate. =A0M= ost of these also have gbean wrappers so they can be accessed through gbean= references.

-- "geronimo" is started from a DS component, EmbeddedDaem= on.

-- I think I'm using the karaf remote jmx security rather than ours. = =A0The capabilities are similar but not identical.

Some other thi= ngs that are not working yet:

-- The (gogo) geronimo console commands that work through "remo= te" gbean proxies don't work AFAIK. =A0Probably one way to fix thi= s would be to expose some more of the DS components using gbean wrappers, b= ut I haven't looked into this yet.

-- the app client (as well as the client-deployer) is not working ye= t at all. =A0We may be able to use command line args to tell the EmbeddedDa= emon it's an app client, or possibly not. =A0We may be able to use a ka= raf instance to supply different ConfigAdmin settings to e.g. the local att= ribute manager to convince it it's an app client. =A0Similarly the sepa= rate console-like things presumably won't work either.

-- the EditableConfigurationManager needs to be replaced by a separa= te component that edits the configuration it gets from the normal configura= tion manager. =A0I think this affects some part of the admin console.

--I couldn't get the xml stream 1.2 and jaxb 2.2 to wo= rk with the spec jars as bundles. =A0=A0=A0According to http://servicemix.396122.n5.nabble.com/= DISCUSS-Enhance-specs-to-work-better-with-JRE-td5001108.html even if yo= u do get them to work (as we seem to have up to now by not exposing the pac= kages from the framework) that breaks other stuff. =A0I think we need to in= vestigate the karaf-activator stuff guillaume wrote and adapt our specs to = use it. =A0At the moment I have the framework lying and claiming later vers= ions for the xmlstream and jaxb packages. =A0I haven't found any docume= ntation for karaf-activator yet.

-- the build uses a lot more memory. =A0I typically run out of permgen = twice during the build with MAVEN_OPTS =3D =A0-XX:MaxPermSize=3D512m -Xms10= 24m -Xmx2048m

-- startup AFAIK only works as ./bin/karaf -l rather t= han our geronimo scripts. =A0Again, I have to increase memory settings for = the server to fully start.


I'f there's no strong oppo= sition I'd like to commit this tomorrow.

Many thanks
david jencks


david jencks




=
--
Thanks!

Regards, Forrest





--
Thanks!
Regards, Forrest




--
Thanks!

Regards, F= orrest




--
Thanks!
Regards, Forrest

--f46d04446c57a6837504b5b0981b--