14:31:22 #startmeeting Karaf Happy Hour 14:31:22 Meeting started Wed Aug 27 14:31:22 2014 UTC. The chair is edwarnicke. Information about MeetBot at http://ci.openstack.org/meetbot.html. 14:31:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:31:22 The meeting name has been set to 'karaf_happy_hour' 14:31:32 #topic Roll Call 14:31:42 #info edwarnicke 14:31:44 #info oflibMichal 14:31:48 #info George Zhao 14:31:49 #info tbachman for Group Based Policy 14:31:52 please #info in :) 14:31:54 #info Phil Robb 14:31:58 #info CASP3R for Integration 14:32:36 #info Rafat for ODL-SDNi App 14:33:51 Anybody else? 14:34:07 2 minute warning for topic switch to Status :) 14:34:08 #info lori for lispflowmapping 14:35:17 #info devinavery (trouble shooting karat feature test failure) 14:36:36 #topic Status 14:36:47 #chair gzhao devinavery tbachman phrobb 14:36:47 Current chairs: devinavery edwarnicke gzhao phrobb tbachman 14:36:54 OK folks... please #info in your status 14:37:26 #info tbachman not very far - editing pom.xml right now in order to run mvn clean install 14:37:35 #info openflowjava: features probably ready, need to discuss a few questions here in this meeting to check correctness 14:37:38 edwarnicke: progress is my own issue, not karaf 14:38:14 Hi! Is this a meeting for Karaf? 14:38:29 hideyuki: yes 14:38:31 hideyuki: yes it is 14:38:45 thanks. 14:38:57 #info edwarnicke adsal-compatibility in on controller, awaiting me providing details in commit message for integration: https://git.opendaylight.org/gerrit/#/c/10350/ 14:39:14 I think POM file (pom.xml) of nsf in controller project lacks dependency. 14:39:53 hideyuki: Which one do you think is missing? 14:40:06 I failed to execute "mvn install" on controller/features/nsf/. 14:40:08 tbachman: Anything I can do to help? 14:40:14 hideyuki: 14:40:16 my work ;) 14:40:16 lol 14:40:21 Let me check here locally with an empty .m2 14:40:23 edwarnicke: feature-base 14:40:25 thx tho 14:40:50 #info lispflowmapping has karaf committed and functional tested 14:41:01 edwarnicke: Yes, after I delete all files under .m2/repository/org/opendaylight, I execute mvn install on controller/features/nsf. 14:41:07 hideyuki: Is the issue in feature-base or feature-nsf? 14:41:09 #info lispflowmapping also submitted its features to integration 14:41:18 lori: Awesome... when can we expect a push to integration? 14:41:22 edwarnicke: the issues in feature-nsf 14:41:27 lori: Could we get a #link ? 14:41:34 hideyuki: Building locally :) 14:41:38 edwarnicke: already done, but the build is unstable 14:41:46 edwarnicke: because an l2switch issue 14:41:52 lori i trigger a new build now 14:41:56 hideyuki: After rm -rf ~/.m2/repository/org/opendaylight/ 14:41:59 edwarnicke: Madhu already proposed a fix for that 14:42:07 not sure if it was merged 14:42:14 CASP3R: lori I think the l2switch issue is fixed 14:42:19 hi...i have one query..On executing feature:list command on karaf container it is displaying my feature but when I am executing bundle:list command it is showing no bundles.Why? 14:42:27 edwarnicke: Yes, after I did rm -rf ~/.m2/repository/org/opendaylight/, i executed mvn install 14:42:31 #link https://git.opendaylight.org/gerrit/#/c/10377/ 14:42:54 #link https://git.opendaylight.org/gerrit/#/c/10364/ <-- lispflowmapping submission for integration 14:42:57 #link https://jenkins.opendaylight.org/integration/job/integration-master-project-centralized-integration/2141/ <- retriggered build 14:43:32 i am working on plugin2oc module and want to integrate karaf with it.. 14:43:53 After I made the following change https://gist.github.com/anonymous/8d5ac78c3b6dcd201238, executing mvn install for nsf-features becomes successful. 14:44:22 hideyuki: Please push a patch :) 14:45:08 edwarnicke: OK. After I push the patch, please review it. I'm not sure my understanding is correct. 14:45:23 hideyuki: Happy to :) 14:45:50 Do we have representatives from yangtools, openflowjava and openflowplugin 14:46:26 gzhao: oflibMichal is from openflowjava 14:46:27 hideyuki: Glancing at your pastebin, you were doing the right *sort* of thing :) 14:46:43 lori: thanks 14:46:51 Jyoti_13: Apologies for taking so long to get to your question 14:47:04 Jyoti_13: are you doing feature:list or feature:list -i 14:47:12 Jyoti_13: In other words, are your features *installed* ? 14:47:14 no problem sir 14:47:22 Jyoti_13: Also, how are you doing your bundle list 14:47:32 sir i am using feature:list 14:47:33 Jyoti_13: One thing to watch out for there, which bites me a lot personally 14:47:46 Jyoti_13: Could you try feature:list -i 14:47:53 Which will show you only the installed ones 14:48:11 yes i tried feature:list -i 14:48:28 Jyoti_13: Also, bundle:list -s is what you need to get a listing of bundles by symbolic name 14:48:34 but there feature was not present 14:48:58 Jyoti_13: Cool, so here's what's going on with your features:list vs features:list -i 14:49:04 Jyoti_13: Your features are *known* 14:49:10 Jyoti_13: But they are not *installed* 14:49:12 try 14:49:17 feature:install 14:49:21 And then see what you can see :) 14:50:37 liemmn: Welcome to the party! :) 14:50:40 i tried installing feature also butit is giving me error somewhat like......package cannot import more than once...something like com.sun.jdmk... 14:50:53 lori: Send me a link to your integration patch and I'll retrigger 14:52:10 @ed :) 14:52:32 gonna work on integration today... 14:52:45 liemmn: Awesome :) 14:52:51 liemmn: Get it into karaf and integration :) 14:53:36 * devinavery raises hand because he doesn't want to have too many questions going on at once... :) 14:53:45 yep, will have more questions later... (I am sure :) ) 14:54:25 edwarnicke: link to IT patch above with #link 14:54:29 devinavery: I can scroll :) Ask at will :) 14:54:36 lori: LOL... will scroll back :) 14:54:42 just a sec 14:54:48 will paste again 14:55:10 edwarnicke: https://git.opendaylight.org/gerrit/#/c/10364/ 14:55:18 lori: It looks like https://git.opendaylight.org/gerrit/#/c/10364/ was retriggered :) 14:55:25 lori: And succeeded :) 14:55:34 i have copied opendaylight-karaf container from controller to my project(which resides under opendaylight).Do I need to make any changes in this karaf container after moving it to my module? 14:55:38 Do we still have CASP3R around? 14:55:40 edwarnicke: hmm, that's VERY recent 14:55:41 edwarnikie: Ok. So I think we have a stale feature reference pointing to an outdated version some where... some quick background: I am trying to build yang tools and controller without access to the ODL repo 14:55:47 I just reloaded 14:55:48 Jyoti_13: Please use the archetype 14:55:49 :) 14:55:55 Jyoti_13: Its much much much better done :) 14:56:02 * tbachman goes into debugging mode... 14:56:08 Jyoti_13: I incorporated a bunch of lessons learned there :) 14:56:13 edwarnikie: when I do this and load up karat some of my odl-netconf features do NOT have descriptions... even though the descriptions are presenting in the features.xml file 14:56:33 devinavery: OK... that does sound wrong 14:56:56 edwarnicke: when I do the same thing on my repo that has access to the public all works well and descriptions are present 14:57:07 Quick poll before I go review lori's integration patch... any not-yet addressed questions hanging out? 14:57:17 devinavery: How far back have you traced it 14:57:23 Note to all, keep the questions coming :) 14:57:33 edwarnicke: I have an observation, not a question 14:57:36 in the archetype 14:57:43 for the features/pom.xml that’s generated 14:57:52 there’s a comment to “move” the properties to the parent pom 14:57:57 I did move them 14:58:02 but got a build error 14:58:09 actually 14:58:09 duh 14:58:15 this is probably my problem 14:58:22 b/c if it were picking up the parent pom 14:58:25 the error wouldn’t happen 14:58:26 nm 14:58:55 edwarnicke: I checked most versions referenced in parent features.xml and pom.xmls etc - all seem to be lining up but clearly something is strange. Is there a way to print an "Effective Features" like there dis the effective pom? 14:58:58 * tbachman just managed to fit in a DoS of 12 lines 14:59:00 tbachman: Please note that is marked as Optional :) 14:59:03 lol 14:59:09 trying to get the A+ 14:59:17 tbachman: I tried to distinguish between 'It would be pretty if you did' and 'shit is going to break if you don't' :) 14:59:26 tbachman: I strongly encourage you in that :) 14:59:35 tbachman: Is your parent pom set correctly in your features/pom.xml 14:59:35 ? 14:59:48 devinavery: Not that I know of 14:59:57 14:59:57 commons.groupbasedpolicy 14:59:59 org.opendaylight.groupbasedpolicy 15:00:00 0.1.0-SNAPSHOT 15:00:01 ../commons/parent 15:00:02 15:00:52 reflecting feature list and bundle list on karaf container is the end output or we need to do something more? 15:01:06 edwarnicke: is there a reason that the config files are separate? 15:01:25 Jyoti_13: Yes... because they get detected and loaded by the config pusher as seprate 15:01:34 tbachman: Hang on, I've got an archetype for that... 15:01:45 * edwarnicke grumbles at himself for not putting it in the docs yet 15:02:22 tbachman: artifact id for config file archetype is opendaylight-configfile-archetype 15:02:33 Just replace opendaylight-karaf-features in the command line 15:02:40 And it will make a nice project for you for your config files 15:02:51 Jyoti_13: What have you got working so far? 15:02:57 edwarnicke: I guess the question is: why do we make the config files separate? 15:03:08 i.e. why not just have them as part of a feature? 15:03:24 would you ever just install a config file? 15:03:48 tbachman: Two reasons 15:04:01 1) What if someone wants a different configuration? 15:04:17 2) We have to detect them coming in and handle them deterministicly 15:04:50 Madhu: Thanks for holding me to account on the adsal-compatibilty stuff at integration, see comments there for reasoning and how I can handle it better next time 15:04:50 i have created pom.xml and feature.xml within features directory of my module which builds successfully and created .kar file which i tried to deploy using karaf. 15:05:10 Jyoti_13: Are you following the instructions here: https://wiki.opendaylight.org/view/Karaf:Step_by_Step_Guide 15:05:36 edwarnicke: that’s sounding like there’s some sort of centralized config file intelligence? 15:05:55 tbachman: The config subsystem insure deterministic behavior 15:05:57 i was reffering to https://meetings.opendaylight.org/opendaylight-meeting/2014/tws/opendaylight-meeting-tws.2014-08-04-17.03.html till now.. 15:06:07 edwarnicke: no, I get that... 15:06:24 tbachman: Basically... when a feature is installed, the config subsytem finds out about it 15:06:33 from yesterday i am following https://wiki.opendaylight.org/view/Karaf:Step_by_Step_Guide 15:06:33 Oh... I think I see what you are getting at 15:06:37 just trying to understand how a user is going to take advantage of this 15:06:39 I can understand that they may want to determine the way a given configuration works 15:06:45 tbachman: You are asking 'couldn't I just attach the config file to an existing bundle' 15:06:47 Yes, you could 15:06:51 yes :) 15:06:54 but 15:06:58 But you will still need the configfile in your feature 15:07:08 As thats how the config subsystem finds out about it 15:07:15 Is that the *right* thing to do 15:07:15 If there’s a good reason otherwise, am fine with it. 15:07:24 right 15:07:26 AAA already has a feature file defined... Can I skip the  opendaylight-karaf-features archetype? 15:07:30 tbachman: My aesthetics would be to keep it separate... but that's just aesthetics, not functionality... and folks are most welcome to have different opinions there :) 15:07:45 liemmn: Is it accessible as a separate file from maven? 15:08:06 lol 15:08:06 Just trying to do what makes sense 15:08:06 if a user can take advantage of this somehow, then sure 15:08:06 yes 15:08:29 liemmn: If you can enter a mvn coordinate for its as a file of type xml, you are good to go 15:08:42 liemmn: See previous comments on aesthetic tastes on my part ;) 15:08:54 liemmn: Apologies 15:08:56 I misread you 15:09:06 ok. 15:09:08 liemmn: So, my suggestion with an existing features.xml 15:09:15 Would be to use the archetype to create a project for you 15:09:17 btw, I tried running the archetype, but it gave me: 15:09:17 [ERROR] The project org.opendaylight.aaa:aaa.project:0.1.0-SNAPSHOT (/home/liemmn/dev/aaa/pom.xml) has 1 error 15:09:18 [ERROR] Child module /home/liemmn/dev/aaa/feature of /home/liemmn/dev/aaa/pom.xml does not exist 15:09:21 Because there are helpful instructions in there 15:09:29 ohh... 15:09:31 nm 15:09:39 And then from there copy in the features you've done your work on 15:09:44 As you follow the instructions 15:09:49 Its not mandatory you do it that way 15:09:59 But you will find it is probably easier to get things right that way 15:10:44 just copying the karaf container into my module and copying the generated .kar file into the deploy folder is OK? 15:10:44 * edwarnicke tries really hard to keep the lines clear between 'You must do things this way or they will break' and 'I suspect you will find it easier if you...' 15:10:55 Jyoti_13: OK in what sense? 15:11:27 Could someone help me? I failed to execute "mvn clean install" on integration/distributions/extra/karaf/ 15:11:40 Jyoti_13: The reasons I ask is that the instructions here: https://wiki.opendaylight.org/view/Karaf:Step_by_Step_Guide would not result in kar files. Now that doesn't mean you couldn't be doing everything right and *also* generating kar files... you could be... but it is a little warning that maybe you *might* be off track 15:11:40 means do I need to make some chnages in karaf container which i have copied? 15:11:49 hideyuki: what is the error ? 15:11:50 hideyuki: I can help, what error are you getting? 15:11:53 I saw the following error. [ERROR] Failed to execute goal on project distribution-karaf: Could not resolve dependencies for project org.opendaylight.integration:distribution-karaf:pom:0.2.0-SNAPSHOT: Failed to collect dependencies for [org.apache.karaf.features:framework:kar:3.0.1 (compile), org.apache.karaf.features:standard:xml:features:3.0.1 (runtime), org.opendaylight.controller:karaf.branding:jar:1.0.0-SNAPSHOT (compile), org.opendayl 15:11:53 ight.controller:opendaylight-karaf-resources:jar:1.4.2-SNAPSHOT (compile), org.opendaylight.integration:features-integration:xml:features:0.2.0-SNAPSHOT (runtime)]: Failed to read artifact descriptor for org.opendaylight.integration:features-integration:xml:features:0.2.0-SNAPSHOT: Could not find artifact org.opendaylight.integration:root:pom:0.2.0-SNAPSHOT in opendaylight-snapshot (http://nexus.opendaylight.org/content/repositories/openda 15:11:58 ylight.snapshot/) -> [Help 1] 15:12:19 https://gist.github.com/anonymous/a5b48cc49be2e48d4ff5 15:12:31 lori: I have not forgotten about you... but and answering live questions right now... 15:12:57 hideyuki: That looks like a local nit in integration 15:13:02 I'm using the latest code of integration Git repository without any chagne. 15:13:04 Build from integration/ and it should work 15:13:13 hideyuki: It looks like integration is not pushing its parent pom.xml to nexus 15:13:25 edwarnicke: Oh, i see. thank you!!!! 15:13:43 hideyuki: Not saying thats right or wrong on integrations part at this time... haven't had a chance to think about it... just saying it is :) 15:14:22 hideyuki: I've hit that error before :) 15:14:26 edwarnicke: another question for you, or any karat expert - where in the karaf distribution do we find the feature.xml files? 15:15:14 cd target/assembly/system 15:15:16 find . 15:15:21 Its laid out like a maven repo 15:15:46 Ok. So when I do that I don't find the "netconf" feature there 15:15:59 In the controller local distro? 15:16:42 edwarnicke: correct. So let me expand on what I am doing. I am trying to confirm that the features-netconf feature.xml is the correct version in my build 15:16:49 Jyoti_13: Still there? Want to make sure you are getting your needs met :) 15:17:03 edwarnicke: when I look in controller/distro.../opendaylight-karaf/target I only see these files: 15:17:10 devinavery: OK... what do you have in your local .m2/repository 15:18:06 hideyuki: Could you add the issue you saw and its fix to a "Common Problems and their solutions" section at the bottom of this section: https://wiki.opendaylight.org/view/Karaf:Step_by_Step_Guide#Prepare_your_feature_for_addition_to_the_integration_features_index 15:18:20 hideyuki: That way the next guy can solve his problem more easily :) 15:18:22 edwarnicke: Madhu recently discovered an issue with jackson 15:18:22 and worked around it 15:18:22 are you aware of the workaround, by chance? 15:18:22 I think it had to do with jersey 15:18:22 like, jersey included something, which was expecting jackson from a certain place 15:18:22 wow 15:18:22 I think I just got a huge burst of IRC :) 15:18:23 edwarnicke: what specifically would you like me to check? I do have a features-netconf-0.2.5-SNAPSHOT-features.xml in my repository 15:18:23 must be a delay on my end 15:18:40 this probably looks very odd to people who don’t have this delay 15:18:43 edwarnicke: Ok. 15:19:07 tbachman: Wait a moment while I try to figure out timelines in my head 15:19:08 tbachman: hi 15:19:13 tbachman: Everything seems so long ago :) 15:19:17 lol 15:19:23 tbachman: Ah... best to let Madhu speak for himself :) 15:19:29 :) 15:19:30 tbachman: yes. i fixed the jackson/jersey issue. 15:19:35 so what is the question ? 15:19:41 * tbachman waits with baited breath 15:19:42 tbachman: Note, odl-restconf uses gson, not jersey 15:19:49 okay i see the question :) 15:19:50 tbachman: Not sure if that detail is relavent 15:19:58 so ... jackson and jersey uses javax.ws.rs 15:20:06 and there are multiple bundles exporting it 15:20:23 but in order for jersey to bind any providers (like json provider for jaxb) 15:20:35 it was expecting the javax.ws.rs exported by jersey.core 15:20:38 if not... 15:20:40 actually till now i have generated .kar file and deploy the same on karaf container... it seems that i have to do rework using https://wiki.opendaylight.org/view/Karaf:Step_by_Step_Guide 15:20:43 the class loader craps out 15:20:54 Jyoti_13: Hopefully not to much rework :) 15:21:28 Jyoti_13: My suggestion would be to move your existing work to one side for refernece, and walk through with the archetypes, as they contain helpful instructions, and copy in your features 15:21:38 but whatever i was doing is that wrong?? 15:21:42 Jyoti_13: Since defining the features is the hard part, hopefully it will go switftly :) 15:21:46 essentially, it was a class name conflict (unable to be detected by human eyes) :) 15:21:46 but runtime can catch it :( 15:21:46 tbachman: is that clear ? 15:21:46 and the fix was to remove the conflicting bundle that exports the javax.ws.rs 15:21:58 Madhu: so, does this require anything special in the features file, or the features/pom.xml? 15:22:05 Jyoti_13: A lot of this has to do with doing things in a way that make sure all features load and can coexist reasonably in the same container 15:22:09 tbachman: in the features file 15:22:10 ok..sure :) 15:22:20 tbachman: lemme show u the gerrit that i used to fix it 15:22:22 1 sec 15:22:24 thx 15:22:38 * tbachman notes that his IRC seems to be incredibly bursty atm 15:22:56 Madhu: Thanks for digging into that, I know such issues are PITA to sort out 15:23:03 like I’ll get about 25 IMs in a second 15:23:08 weird 15:23:08 (and of course about 40 seconds of dead air) 15:23:26 edwarnicke: do our feature unit tests etc rely on hardcoded repository locations...? 15:23:47 tbachman: https://git.opendaylight.org/gerrit/#/c/10111/5/features/base/src/main/resources/features.xml 15:23:52 you can see that i removed a reference to javax.ws.rs 15:23:56 Madhu: thx! 15:24:09 tbachman: So is mine 15:24:21 edwarnicke: weird, eh? 15:24:27 edwarnicke: or even our karaf distributions? 15:24:39 either that, or you all are *amazingly* fast typists :) 15:24:58 devinavery: They shouldn't rely on hardcoded repository locations, but they do rely on the dependencies being in the local .m2/repository (or in maven.repo.local if specified) 15:25:07 tbachman: I have *really* good coffee 15:25:10 lol 15:25:55 edwarnicke: ok. We are overriding the repo location with maven.repo.local so I wiped out my old ~/.m2/repository just in case and am retrying 15:26:05 Guys... 5 minute warning... the pre-M5 meeting happens here in 5 minutes 15:26:10 btw 15:26:12 pre-M5 coming up 15:26:22 Guys I suspect we can keep chatting karaf, but will leave that call to phrobb 15:26:36 I suspect we at minimum though should shut down our meeting and let gzhao or phrobb start up the pre-M5 meeting 15:26:41 edwarnicke: would think M5 might trump 15:26:49 * tbachman agrees 15:26:55 we could move to #opendaylight? 15:26:57 devinavery: I *know* using -Dmaven.repo.local works... else Jenkins would die 15:27:05 tbachman: Hmm... thinking 15:27:10 edwarnicke: ok. Thanks 15:27:24 phrobb: gzhao how would you guys like to handle things (assuming you are organizing the M5 meeting) 15:27:44 Does GUI work on Karaf distribution? 15:27:54 devinavery: Please note, the probablility that the magic variable is actually named what I typed above is low... but whatever it is, it *does* work :) 15:28:04 hideyuki: I think dlux is going in momentarily 15:28:20 hideyuki: When I went to sleep last night, I believe mlemay had it verifying 15:28:23 edwarnicke: in Doc meeting with Matheiu and Phil now 15:28:24 mlemay: Where are we with dlux 15:28:31 edwarnicke: multi-booked 15:28:36 MEETING MANIA! :) 15:29:04 gzhao: What IRC channel... need to ask mlemay somthing :) 15:29:10 edwarnicke: I see. 15:29:18 edwarnicke: yes, need to switch the pre M5 meeting 15:29:41 gzhao: Should I #endmeeting now then? 15:30:47 edwarnicke: Yes, we need to start the M5 Developer meeting. 15:30:50 edwarnicke: understood 15:30:58 #endmeeting