| time |
nick |
message |
08:32 |
<gertv> |
gnodet: ever seen SMX4 stall at startup -- very early on in the process? |
08:34 |
<gertv> |
edelln: will you be providing additional patches for moving the other examples or do you want me to do that? |
08:36 |
<edelln> |
10gertv:01 I have them almost complete - I will just update the jbi maven plugin to test them and supply you a patch |
08:37 |
<gertv> |
edelln: ok, thx once again for the patch |
08:41 |
<ffang> |
gertv: I'v seen smx4 hang intermittently when start it, but it's not very often, I'm on Linux |
08:42 |
<gertv> |
ffang: yeah, me too |
08:44 |
<gertv> |
it doesn't ever do it the first time you start it, but if you 'exit' from the console and then try again, you're bound to run into it at some point |
08:46 |
<ffang> |
gertv: yeah, sometimes it happen even with empty data folder |
08:48 |
<ffang> |
gertv: we may need raise a jira to track this issue |
08:48 |
<gertv> |
ffang: yeah, will do |
11:36 |
<gertv> |
gnodet: could you review http://svn.eu.apache.org/viewvc?view=rev&revision=677248 for me? |
11:37 |
<gertv> |
gnodet: did you intend to protect the factories map from concurrent access with the synchronized on bundleChanged? |
11:37 |
<gnodet> |
gertv: is this supposed to solve the problem where smx4 hangs at startup ? |
11:37 |
<gnodet> |
yes, i think so |
11:38 |
<gertv> |
yes, I think it does solve it |
11:39 |
<gertv> |
gnodet: the bundleChanged method was called while still in the start() method -- causing a lock on the Activator object |
11:40 |
<gnodet> |
gertv: but if this is the case, the bundleContext will not be set correctly at the time the bundleChanged() method is called ? |
11:41 |
<gnodet> |
which should not be a problem, as it is not used but in the stop method it seems |
11:41 |
<gertv> |
why not? it is set before even adding the listener |
11:43 |
<gnodet> |
i don't follow |
11:44 |
<gnodet> |
bundleChanged was called by the osgi framework while still in the start() metho, right ? |
11:44 |
<gnodet> |
this is because the iteration in start() takes a long time |
11:45 |
<gertv> |
yes, but if you first set the bundleContext and only then add the class as a bundle listener |
11:45 |
<gnodet> |
good point |
11:45 |
<gertv> |
it shouldn't get bundleChanged() callback before it has been set a bundle listener |
11:45 |
<gnodet> |
it looks good then |
11:46 |
<gnodet> |
meaning we have to re-release the specs ... |
11:46 |
<gertv> |
yep |
11:47 |
<gertv> |
thanks for reviewing, always appreciate a second opinion on these type of locking/concurrency issues |
11:47 |
<gertv> |
they're so easy to mess up |
12:13 |
<edelln> |
I have the examples ready to submit for smx4 but when I am building a kit just from features trunk it attempts to copy the jar file into a directory in my local repo "SU_NAME.jbi-service-unit and so when it trys to find the dependencies for the assemblies it falls over |
12:14 |
<edelln> |
if I go and build examples beforehand all the jars will be in my local repo |
12:15 |
<gertv> |
edelln: not sure I get what the problem is |
12:15 |
<gertv> |
the assembly should only copy over the sources and readme/pom, no? |
12:16 |
<edelln> |
Well this is when I do mvn install from the top of the features project |
12:16 |
<edelln> |
so its supposed to build examples before assembly |
12:17 |
<gertv> |
ok |
12:17 |
<edelln> |
If you delete your examples directory in your local repo and try this it will fall over when building the bridge |
12:17 |
<gertv> |
but it doesn't do it in that order? |
12:18 |
<gertv> |
gertv is giving it a try |
12:18 |
<edelln> |
It does but I never see the jar files for the service units been copied to my local repo - it just copies the zips |
12:27 |
<gertv> |
gertv needs to rebuild the maven jbi plugin again as well |
12:31 |
<gertv> |
edelln: ok, I see your problem now |
12:33 |
<edelln> |
10gertv:01 yeah and if I build the examples locally to get around this issue - it then falls over with 10Unrecognised tag: 'filtered'01 |
12:34 |
<edelln> |
I guess this is due to the maven plugin updates |
12:34 |
<gertv> |
edelln: yes, the 4.0-SNAPSHOT version of the plugin is broken |
12:35 |
<gertv> |
it might have been before as well - smx 3.3 branch uses the released 3.2.1 version |
12:36 |
<gertv> |
could you raise yet another issue for it and I'll take a look as soon as I got these NPE from ESB-321 fixed? |
12:36 |
<edelln> |
yep! |
12:37 |
<edelln> |
I will put the patch together anyway as I have changed the version attribute to be a property in all the demos so when it does change we only need to change it in one place |
12:38 |
<gertv> |
great! we should be able to release it with plugin version 3.2.1 as well if I can't get it fixed in time anymore |
12:46 |
<php> |
سلام |
12:47 |
<edelln> |
10gertv: 01https://issues.apache.org/activemq/browse/SMX4-64 |
12:47 |
<php> |
hi |
12:58 |
<edelln> |
10 gertv: 01 I added the demos and patch to https://issues.apache.org/activemq/browse/SMX4-61 so when you get a chance if you can apply them that would be great |
12:58 |
<gertv> |
edelln: I've just fixed the NPE thingy, so I'll do it right away |
12:58 |
<gertv> |
edelln: thx for the patch |
12:59 |
<edelln> |
great! |
14:27 |
<gnodet> |
gertv: are you looking at SMX4-64 ? |
14:28 |
<gertv> |
yeah, I was just going to ping you for that |
14:29 |
<gertv> |
gnodet: very strange, doesn't happen if you start the build from the /examples directory |
14:29 |
<gertv> |
but it does fail if you start it from features root |
14:30 |
<gertv> |
does that ring any bells? |
14:33 |
<gnodet> |
not yet |
14:34 |
<gertv> |
gnodet: I can only think of some dependency being pulled in in wrong version by a plugin defined at the root pom.xml |
14:43 |
<gertv> |
gnodet: but that isn't the case -- just ran both of them redirecting output and I can't see the difference there |
14:45 |
<gertv> |
gnodet: how does it know to use .jar? only from the ArtifactHandler registered in plexus/components.xml? |
14:46 |
<gnodet> |
can you paste the log, i'm still downloading tons of stuff while building |
14:47 |
<gertv> |
those things are huge as well -- 8 MB |
14:47 |
<gertv> |
ran mvn with -X to get more info |
14:47 |
<gertv> |
what do you want to see? or do you want me to send them over to you? |
14:48 |
<gnodet> |
what is the ".jar" you're talking about ? |
14:49 |
<gertv> |
well, if you run the thing from features root, it does |
14:49 |
<gertv> |
[INFO] Installing /home/gert/Projects/smx4/features/examples/bridge/bridge-http-su/target/bridge-http-su-4.0-m2-SNAPSHOT.jar to /home/gert/.m2/repository/org/apache/servicemix/examples/bridge/bridge-http-su/4.0-m2-SNAPSHOT/bridge-http-su-4.0-m2-SNAPSHOT.jbi-service-unit |
14:49 |
<gertv> |
[INFO] Installing /home/gert/Projects/smx4/features/examples/bridge/bridge-http-su/target/bridge-http-su-4.0-m2-SNAPSHOT.zip to /home/gert/.m2/repository/org/apache/servicemix/examples/bridge/bridge-http-su/4.0-m2-SNAPSHOT/bridge-http-su-4.0-m2-SNAPSHOT.zip |
14:49 |
<gertv> |
notice the .jbi-service-unit extension |
14:50 |
<gertv> |
running it from examples yields |
14:50 |
<gertv> |
[INFO] [install:install] |
14:50 |
<gertv> |
[INFO] Installing /home/gert/Projects/smx4/features/examples/bridge/bridge-http-su/target/bridge-http-su-4.0-m2-SNAPSHOT.jar to /home/gert/.m2/repository/org/apache/servicemix/examples/bridge/bridge-http-su/4.0-m2-SNAPSHOT/bridge-http-su-4.0-m2-SNAPSHOT.jar |
14:50 |
<gertv> |
[INFO] Installing /home/gert/Projects/smx4/features/examples/bridge/bridge-http-su/target/bridge-http-su-4.0-m2-SNAPSHOT.zip to /home/gert/.m2/repository/org/apache/servicemix/examples/bridge/bridge-http-su/4.0-m2-SNAPSHOT/bridge-http-su-4.0-m2-SNAPSHOT.zip |
14:50 |
<gnodet> |
sounds like the maven plugin extensions are not loaded ? |
14:51 |
<gnodet> |
have you tried adding a pluginManagement section for the jbi maven plugin in the root pom ? |
14:51 |
<gertv> |
yep |
14:51 |
<gertv> |
still no good |
14:52 |
<gnodet> |
with the <extension>true</extensions> tag ? |
14:53 |
<gertv> |
<plugin> |
14:53 |
<gertv> |
<groupId>org.apache.servicemix.tooling</groupId> |
14:53 |
<gertv> |
<artifactId>jbi-maven-plugin</artifactId> |
14:53 |
<gertv> |
<version>${servicemix-jbi-plugin-version}</version> |
14:53 |
<gertv> |
<extensions>true</extensions> |
14:53 |
<gertv> |
</plugin> |
14:58 |
<gertv> |
well, there are these org.codehaus.plexus.component.repository.exception.ComponentLookupException: Component descriptor cannot be found in the component repository: org.apache.maven.lifecycle.mapping.LifecycleMappingjbi-service-unit. |
14:58 |
<gertv> |
|
14:58 |
<gertv> |
in the log file as well |
14:59 |
<gertv> |
but they appear for both the failing and succeeding builds |
15:10 |
<gertv> |
gnodet: no science involved, just 'flying by google', but... |
15:10 |
<gertv> |
* @parameter expression="${project.build.directory}/classes" |
15:10 |
<gertv> |
*/ |
15:10 |
<gertv> |
private File serviceUnitLocation; |
15:10 |
<gertv> |
+ |
15:10 |
<gertv> |
+ /** |
15:10 |
<gertv> |
+ * @parameter expression="${component.org.apache.maven.artifact.handler.ArtifactHandler#jbi-service-unit}" |
15:10 |
<gertv> |
+ */ |
15:10 |
<gertv> |
+ private ArtifactHandler handler; |
15:10 |
<gertv> |
|
15:10 |
<gertv> |
public void execute() throws MojoExecutionException, MojoFailureException { |
15:10 |
<gertv> |
try { |
15:10 |
<gertv> |
- |
15:11 |
<gertv> |
+ project.getArtifact().setArtifactHandler(handler); |
15:11 |
<gertv> |
createUnpackedInstaller(); |
15:11 |
<gertv> |
|
15:11 |
<gertv> |
File serviceUnitFile = new File(outputDirectory, serviceUnitName); |
15:11 |
<gertv> |
this seems to fix it on my machine |
15:11 |
<gnodet> |
sounds good, but why would it only appear now ? |
15:11 |
<gnodet> |
is there any change in maven that cause the problem ? |
15:12 |
<gertv> |
like I said: no science involved -- I don't have the slightest idea |
15:14 |
<gnodet> |
ok, so you don't run into the IncompatibleClassChangeError if you use 4.0-SNAPSHOT ? |
15:14 |
<gertv> |
no, I have fixed that yesterday |
15:14 |
<gertv> |
but you do need to rebuild plugins for that |
15:14 |
<gnodet> |
ah, haven't svn up the plugins |
15:15 |
<gertv> |
and they haven't been rebuilt yet at EC2 either |
15:15 |
<gertv> |
gnodet: do you want to dig deeper to find the root cause for the ArtifactHandler? |
15:15 |
<gnodet> |
no |
15:16 |
<gertv> |
ok, so I just commit my 'thx-to-google-fix' and get it over with |
15:16 |
<gnodet> |
i was just curious, but i don't think it's worth spending too much time on that |
15:16 |
<gnodet> |
yeah |
15:20 |
<gnodet> |
gertv: is there anyhting left to do on SM-1455 ? all components / shared libs depend on 3.2.1 now |
15:20 |
<gnodet> |
or did you meant something else ? |
15:21 |
<gertv> |
should they depend on that for anything but testing? |
15:21 |
<gertv> |
the JBI API perhaps? |
15:23 |
<gnodet> |
well, we can't remove all the dependencies right now |
15:23 |
<gertv> |
gnodet: could you do an svn up on the plugin to check it on your machine as well? |
15:23 |
<gnodet> |
for example, they depend on the jbi.jaxp package |
15:23 |
<gnodet> |
it has been extracted out of smx-core, but not released yet |
15:23 |
<gnodet> |
so either we depend on smx-core-3.2.1, or smx-utils-3.3-SNAPSHOT |
15:24 |
<gertv> |
but of them are part of the container, right? |
15:24 |
<gnodet> |
there are still a few dependencies, right |
15:24 |
<gnodet> |
that's why smx-core is included in servicemix-shared atm |
15:25 |
<gertv> |
shouldn't these things go outside of containers and components in the long run? |
15:25 |
<gertv> |
like a servicemix-commons project to hold them? |
15:25 |
<gnodet> |
well, maybe, but the problem is that parts of the container use those packages |
15:25 |
<gnodet> |
so we'd really have to prune things like lightweight components |
15:26 |
<gnodet> |
to get rid of most of the problem i think |
15:26 |
<gertv> |
but if we separate them out in e.g. servicemix-commons and release that |
15:26 |
<gnodet> |
i haven't done a thorough analysis of the dependencies though |
15:26 |
<gertv> |
both the components and the container can depend on that |
15:27 |
<gertv> |
and we can reuse it for smx4 as well |
15:27 |
<gertv> |
would make sense for things like JAXP which is all about custom Source implementation and stuff like that iirc |
15:27 |
<gnodet> |
yeah, it should be possible now that we don't depend on 3.3-SNAPSHOT anymore |
15:28 |
<gertv> |
for itesting, we still have to rely on the latest container release |
15:30 |
<gertv> |
gnodet: btw, the thesis student that asked about monitoring a few weeks ago is going to do his work on the JCR auditing stuff in sandbox |
15:30 |
<gnodet> |
yay! nice :-) |
15:35 |
<edelln> |
guys! I rebuilt the plugins and tryed to build the kit again but I still get the same issue (windows) |
15:38 |
<gnodet> |
edelln: did you change the version of the plugin used in the root pom.xml ? |
15:39 |
<gnodet> |
afaik, it points to 3.2.1 and not 4.0-SNASPHOT |
15:40 |
<gertv> |
well, I did add the plugin the pluginmanagement here as well |
15:40 |
<gertv> |
let me commit that too so you can give it another go |
15:41 |
<edelln> |
yep its 3.2.1 as I was trying to play around fixing this as well |
15:42 |
<gertv> |
edelln: if you svn up, everything should be there |
15:42 |
<edelln> |
ok will give it another go |
15:46 |
<gertv> |
edelln: I do get the exception about the <filtered> element now for the first time? did you perhaps mean that one? |
15:46 |
<gertv> |
I just see now that it's in the same issue as the .jbi-service-unit thing |
15:47 |
<edelln> |
yeah! I hit that after once I had the demos builts locally |
15:48 |
<edelln> |
I thought they were related so put them in the one issue |
15:48 |
<gertv> |
no, they're not, but sorry I missed it |
15:49 |
<gertv> |
gertv promises to read the *entire* issue next time |
15:50 |
<gertv> |
that's probably because it picked up an older version of the assembly plugin somewhere/somehow |
15:52 |
<edelln> |
ok let me check the demos as I put a property in the root pom for the version of the assembly plugin - maybe I missed one but they all seem to be using 2.1 |
15:52 |
<gertv> |
that should be 2.2-beta-2 instead of 2.1 |
15:52 |
<gertv> |
we need it to handle snapshot and filtering correctly |
15:53 |
<edelln> |
yeah alot of the demos had 2.1 hardcode in their poms |
15:53 |
<gertv> |
ok, let me try to fix that |
15:58 |
<edelln> |
I did catch all the hardcoded versions so all you should have to change is the property in the root pom |
16:02 |
<gertv> |
could you svn up and give it another try? |
16:02 |
<edelln> |
sure |
16:03 |
<gertv> |
jstrachan: could you run the servicemix maven plugins build manually on the CI build box? |
16:07 |
<jstrachan> |
sure - is it not run now? |
16:08 |
<gnodet> |
gertv: can you check if https://issues.apache.org/activemq/browse/SM-1396 is solved now ? i think that's the issue you fixed yesterday, right ? |
16:09 |
<gnodet> |
or maybe this is unrelated, if so i'll commit the provided patch |
16:10 |
<gertv> |
this is unrelated, I think |
16:11 |
<gnodet> |
ok |
16:12 |
<gertv> |
jstrachan: it is run, but only after the SMX4 build |
16:12 |
<gertv> |
and the SMX4 build is going to fail unless we have an updated snapshot version of the plugins first |
16:13 |
<jstrachan> |
smx3 or smx4? |
16:13 |
<jstrachan> |
can't we just hack the smx[3|4] shell script to change the order? |
16:13 |
<gertv> |
I looked at that, but it's a different shell scripts |
16:13 |
<jstrachan> |
ah ok |
16:14 |
<jstrachan> |
jstrachan wishes he'd figured out how to put the crontab into svn |
16:14 |
<gertv> |
the one that needs to be run is smx-mvn-plugins-pom.sh |
16:14 |
<jstrachan> |
thats run before smx4-nightly apparenlty |
16:14 |
<gertv> |
ok, so I misread those wacky american time zone then |
16:14 |
<jstrachan> |
current crontab: http://rafb.net/p/wI0pRj93.html |
16:15 |
<gertv> |
ok, thx for looking into it then |
16:19 |
<edelln> |
yay! I got the kit built |
16:20 |
<gertv> |
edelln: yay! |