I have forgotten to mention the problem occurs on vanilla Karaf too.
On 11.06.2014 11:19, Sobkowiak, Krzysztof wrote: > I have noticed something interesting with the quickstart routes (even > with Karaf 2.3.4, but it doesn't seem to cause our problem). When > Karaf starts you can see in the log the routes are started, next shut > down and started again. Probably something connected to SM-2325. With > Karaf 2.3.4 the second route startup is successful, with Karaf 2.3.5+ > the bundle stays in status Creating (immediately with 2.3.5 or after > more restarts with 2.3.6). > > > On 11.06.2014 11:11, Sobkowiak, Krzysztof wrote: >> No. I haven't created it yet (to less time last weeks). You'll >> probably better decide where to open this issue. >> >> Best regards >> Krzysztof >> >> On 11.06.2014 11:09, Jean-Baptiste Onofré wrote: >>> By the way, it could be related to the Aries Blueprint update (I >>> have to double check). >>> >>> Regards >>> JB >>> >>> On 06/11/2014 11:05 AM, Sobkowiak, Krzysztof wrote: >>>> Hi Jean-Baptiste >>>> >>>> There is one more issue for hot deployment (SM-2325). It is rather a >>>> Karaf theme (I could reproduce it on Karaf) but I think it could have >>>> an impact for hot deployment of Camel routes too. It looks like the >>>> route once installed by hot deployment was started while Karaf startup >>>> and next the deployer detected the blueprint file in deploy directory >>>> and tried to update/refresh the already deployed blueprint. But you >>>> will >>>> probably better know what can cause the problem (or if this is >>>> really a >>>> problem) >>>> >>>> Best regards >>>> Krzysztof >>>> >>>> On 11.06.2014 10:52, Jean-Baptiste Onofré wrote: >>>>> Hi guys, >>>>> >>>>> I reopen the Jira this morning to investigate and I plan to work on >>>>> the Karaf 2.3.6 Jira this afternoon and tomorrow. >>>>> It's a question of timing. I should have close the Jira for a Karaf >>>>> 2.3.6 release by the end of this week. So Karaf 2.3.6 could be out >>>>> next week. >>>>> >>>>> If we consider it's a bit late, we can release ServiceMix 5.1.0/5.0.2 >>>>> with Karaf 2.3.4 (and updated Camel/CXF/...) today and prepare >>>>> ServiceMix 5.1.1/5.0.3 in 2/3 weeks. >>>>> >>>>> Regards >>>>> JB >>>>> >>>>> On 06/11/2014 10:47 AM, Gert Vanthienen wrote: >>>>>> Hi Krzysztof, >>>>>> >>>>>> That's not good news, indeed! Can you perhaps take a thread dump >>>>>> next >>>>>> time you see this, it might help us figure out why it's going >>>>>> wrong or >>>>>> what the best way to fix it is. >>>>>> >>>>>> We are now almost 10 days after the first attempt to release 5.1.0. >>>>>> Even if we can fix this in Karaf today and start a release vote >>>>>> there, >>>>>> it would still be early next week at the very best before we can >>>>>> build >>>>>> another release candidate for ServiceMix. Would it worth >>>>>> considering >>>>>> to do a ServiceMix 5.1.0 and 5.0.2 release with Karaf 2.3.4 (so at >>>>>> least users would have the updated versions of Camel, Activiti, ... >>>>>> available) and then follow up with 5.1.1/5.0.3 once Karaf 2.3.6 is >>>>>> out? >>>>>> >>>>>> Regards, >>>>>> >>>>>> Gert Vanthienen >>>>>> >>>>>> >>>>>> On Thu, Jun 5, 2014 at 10:46 PM, Krzysztof Sobkowiak >>>>>> <[email protected]> wrote: >>>>>>> Hi >>>>>>> >>>>>>> I have bad news. I have just tested the problem again using the new >>>>>>> 2.3.6-SNAPSHOT and the problems still exists (but after more, 3-4, >>>>>>> restarts). I have reverted the changes for KARAF-2760 and built >>>>>>> Karaf. >>>>>>> It solved the problem. It means, KARAF-2760 causes the problem. >>>>>>> But it >>>>>>> seems KARAF-2910 causes the problem does not exist after first >>>>>>> restart. >>>>>>> But after more restarts the problem exists again. >>>>>>> >>>>>>> Best regards >>>>>>> Krzysztof >>>>>>> >>>>>>> On 01.06.2014 19:16, Jean-Baptiste Onofré wrote: >>>>>>>> I agree, but please, give a couple of days to update and integrate >>>>>>>> some fixes in 2.3.6-SNAPSHOT. I will tackle the update in >>>>>>>> ServiceMix >>>>>>>> myself if you don't mind. >>>>>>>> >>>>>>>> Thanks, >>>>>>>> Regards >>>>>>>> JB >>>>>>>> >>>>>>>> On 06/01/2014 07:05 PM, Krzysztof Sobkowiak wrote: >>>>>>>>> +1 for reverting to 5.0.0 (as 5.0.1 is not usable) and releasing >>>>>>>>> 5.0.2 >>>>>>>>> based om Karaf 2.3.6 >>>>>>>>> +1 to releasing 5.1.0 with Karaf 2.3.6 >>>>>>>>> Perhaps could we change now the code base to Karaf >>>>>>>>> 2.3.6-SNAPSHOT to >>>>>>>>> have more time for tests. >>>>>>>>> >>>>>>>>> >>>>>>>>> Best regards >>>>>>>>> Krzysztof >>>>>>>>> >>>>>>>>> On 31.05.2014 11:40, Gert Vanthienen wrote: >>>>>>>>>> L.S., >>>>>>>>>> >>>>>>>>>> Looking at the changes for that issue, that change would >>>>>>>>>> definitely >>>>>>>>>> avoid the issue I was seeing on my machine. I'll ping the Karaf >>>>>>>>>> dev@ >>>>>>>>>> list to see if we can get a new 2.3.6 release out soon enough >>>>>>>>>> for us >>>>>>>>>> to work with. If that would take too long, I think I would >>>>>>>>>> prefer >>>>>>>>>> doing a 5.1.0 with an older version of Karaf, but providing the >>>>>>>>>> 2.13.1 >>>>>>>>>> version of Camel to our users instead of waiting around - we can >>>>>>>>>> always do a 5.1.1 afterwards when Karaf 2.3.6 is out. >>>>>>>>>> >>>>>>>>>> In the meanwhile, I'm wondering if we shouldn't revert to >>>>>>>>>> promoting >>>>>>>>>> 5.0.0 on our main website/downloads or at least add a "known >>>>>>>>>> issues" >>>>>>>>>> section to the release notes about the possible ActiveMQ issue? >>>>>>>>>> >>>>>>>>>> Regards, >>>>>>>>>> >>>>>>>>>> Gert >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Fri, May 30, 2014 at 10:01 PM, Krzysztof Sobkowiak >>>>>>>>>> <[email protected]> wrote: >>>>>>>>>>> Hi >>>>>>>>>>> >>>>>>>>>>> I don't know if it helps, but I have tested the same problem on >>>>>>>>>>> Karaf >>>>>>>>>>> snapshots (4.0.0-SNAPSHOT, 3.0.2-SNAPSHOT, 2.3.6-SNAPSHOT) >>>>>>>>>>> and the >>>>>>>>>>> problem seems to be fixed. Has fix for issue >>>>>>>>>>> https://issues.apache.org/jira/browse/KARAF-2910 fixed the >>>>>>>>>>> problem (if >>>>>>>>>>> it's really caused by issue mentioned by Gert)? >>>>>>>>>>> >>>>>>>>>>> Of course, we are still on 2.3.5 and 3.0.1. >>>>>>>>>>> >>>>>>>>>>> Best regards >>>>>>>>>>> Krzysztof >>>>>>>>>>> >>>>>>>>>>> On 30.05.2014 18:05, Jean-Baptiste Onofré wrote: >>>>>>>>>>>> It's for me ;) >>>>>>>>>>>> >>>>>>>>>>>> Let me take a look on that and especially the default >>>>>>>>>>>> config in >>>>>>>>>>>> etc/org.apache.karaf.features.cfg: I introduced a config to >>>>>>>>>>>> control of >>>>>>>>>>>> the feature (sync or async), it could be related (or the start >>>>>>>>>>>> level >>>>>>>>>>>> support). However, the default behavior should be the same. >>>>>>>>>>>> >>>>>>>>>>>> Let me take a look on that. >>>>>>>>>>>> >>>>>>>>>>>> I keep you posted. >>>>>>>>>>>> >>>>>>>>>>>> Regards >>>>>>>>>>>> JB >>>>>>>>>>>> >>>>>>>>>>>> On 05/30/2014 06:02 PM, Krzysztof Sobkowiak wrote: >>>>>>>>>>>>> I have tested this problem wit AMQ 5.9.0 and 5.9.1 on >>>>>>>>>>>>> Karaf 2.3.4 >>>>>>>>>>>>> and >>>>>>>>>>>>> 2.3.5. On 2.3.4 it works, on 2.3.5 start of the routes fails >>>>>>>>>>>>> after Karaf >>>>>>>>>>>>> restart. It seems to be a problem with Karaf upgrade. >>>>>>>>>>>>> >>>>>>>>>>>>> Best regards >>>>>>>>>>>>> Krzysztof >>>>>>>>>>>>> >>>>>>>>>>>>> On 30.05.2014 16:51, Gert Vanthienen wrote: >>>>>>>>>>>>>> L.S., >>>>>>>>>>>>>> >>>>>>>>>>>>>> I took a quick look at the issue with the ActiveMQ >>>>>>>>>>>>>> connection >>>>>>>>>>>>>> factory >>>>>>>>>>>>>> that is keeping the quickstart example in "Creating" - on my >>>>>>>>>>>>>> machine, >>>>>>>>>>>>>> I also have the same issue with Apache ServiceMix 5.0.1 >>>>>>>>>>>>>> (which is >>>>>>>>>>>>>> still using ActiveMQ 5.9.0). Looking at the output from >>>>>>>>>>>>>> jstack, it >>>>>>>>>>>>>> looks like the changes for >>>>>>>>>>>>>> https://issues.apache.org/jira/browse/KARAF-2760 might have >>>>>>>>>>>>>> something >>>>>>>>>>>>>> to do with it - when the properties LoginModule is being >>>>>>>>>>>>>> initialized >>>>>>>>>>>>>> as part of the ActiveMQ connection being set up, it wants to >>>>>>>>>>>>>> register >>>>>>>>>>>>>> something in the OSGi Service Registry and seems unable to >>>>>>>>>>>>>> acquire the >>>>>>>>>>>>>> necessary locks to do that. >>>>>>>>>>>>>> >>>>>>>>>>>>>> For commons-lang: that seems to be caused by a change in the >>>>>>>>>>>>>> camel-cxf >>>>>>>>>>>>>> feature definition. In earlier versions, that was >>>>>>>>>>>>>> installing >>>>>>>>>>>>>> the >>>>>>>>>>>>>> cxf-xjc-runtime feature which included the commons-lang >>>>>>>>>>>>>> bundle. >>>>>>>>>>>>>> With >>>>>>>>>>>>>> the new versions, a smaller set of CXF features is being >>>>>>>>>>>>>> installed as >>>>>>>>>>>>>> part of the camel-cxf feature, causing the commons-lang >>>>>>>>>>>>>> to no >>>>>>>>>>>>>> longer >>>>>>>>>>>>>> be installed out-of-the-box. So that seems to be an >>>>>>>>>>>>>> intended >>>>>>>>>>>>>> change >>>>>>>>>>>>>> in the Camel features.xml file. Since none of the remaining >>>>>>>>>>>>>> boot >>>>>>>>>>>>>> features require commons-lang, I don't think there's a >>>>>>>>>>>>>> real need >>>>>>>>>>>>>> for >>>>>>>>>>>>>> it to be installed by default. >>>>>>>>>>>>>> >>>>>>>>>>>>>> Regards, >>>>>>>>>>>>>> >>>>>>>>>>>>>> Gert Vanthienen >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> On Fri, May 30, 2014 at 3:10 PM, Gregor Zurowski >>>>>>>>>>>>>> <[email protected]> wrote: >>>>>>>>>>>>>>> Hi Everyone: >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> I have found similar issues related to ActiveMQ as >>>>>>>>>>>>>>> previously >>>>>>>>>>>>>>> reported >>>>>>>>>>>>>>> by Krzysztof and Jean-Baptiste: Custom bundles using >>>>>>>>>>>>>>> ActiveMQ >>>>>>>>>>>>>>> did not >>>>>>>>>>>>>>> reflect the correct status after deployment, and the >>>>>>>>>>>>>>> broker did >>>>>>>>>>>>>>> not >>>>>>>>>>>>>>> get started with another test installation. >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> I had further problems with a custom bundle originally >>>>>>>>>>>>>>> developed for >>>>>>>>>>>>>>> 5.0.0 because the org.apache.commons.lang bundle is not >>>>>>>>>>>>>>> available by >>>>>>>>>>>>>>> default any longer. Further looking into it, I realized >>>>>>>>>>>>>>> that >>>>>>>>>>>>>>> this >>>>>>>>>>>>>>> seems to be the case since 5.0.1. Is that an intended >>>>>>>>>>>>>>> change >>>>>>>>>>>>>>> or an >>>>>>>>>>>>>>> actual defect? >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Thanks, >>>>>>>>>>>>>>> Gregor >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> On Wed, May 28, 2014 at 8:29 AM, Gert Vanthienen >>>>>>>>>>>>>>> <[email protected]> wrote: >>>>>>>>>>>>>>>> L.S., >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> This is a vote to release Apache ServiceMix 5.1.0. >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> The staging area is available at >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> https://repository.apache.org/content/repositories/orgapacheservicemix-1009/ >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> An overview of issues fixed in this release can be >>>>>>>>>>>>>>>> found in >>>>>>>>>>>>>>>> JIRA at >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> https://issues.apache.org/jira/browse/SM/fixforversion/12326660 >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> The scm tag is available on >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> https://git-wip-us.apache.org/repos/asf/servicemix/?p=servicemix.git;a=commit;h=8504e67e8eaffca62aae15936daa85c5e82e53a5 >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Please vote to approve this release: >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> [ ] +1 Approve the release >>>>>>>>>>>>>>>> [ ] -1 Do not approve the release (please provide specific >>>>>>>>>>>>>>>> comments) >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> This vote will be open for 72 hours. >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Regards, >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Gert Vanthienen >>>>>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> Krzysztof Sobkowiak >>>>>>> >>>>>>> JEE & OSS Architect | Technical Architect @ Capgemini >>>>>>> Capgemini <http://www.pl.capgemini.com/> | Software Solutions >>>>>>> Center >>>>>>> <http://www.pl.capgemini-sdm.com/> | Wroclaw >>>>>>> e-mail: [email protected] >>>>>>> <mailto:[email protected]> | >>>>>>> Twitter: @KSobkowiak >>>>>>> Calendar: goo.gl/yvsebC >>>>> >>>> >>> >> >> -- >> Krzysztof Sobkowiak >> >> JEE & OSS Architect | Technical Architect @ Capgemini >> Capgemini <http://www.pl.capgemini.com/> | Software Solutions Center >> <http://www.pl.capgemini-sdm.com/> | Wroclaw >> e-mail: [email protected] <mailto:[email protected]> >> | Twitter: @KSobkowiak >> Calendar: http://goo.gl/yvsebC > > -- > Krzysztof Sobkowiak > > JEE & OSS Architect | Technical Architect @ Capgemini > Capgemini <http://www.pl.capgemini.com/> | Software Solutions Center > <http://www.pl.capgemini-sdm.com/> | Wroclaw > e-mail: [email protected] <mailto:[email protected]> | > Twitter: @KSobkowiak > Calendar: http://goo.gl/yvsebC -- Krzysztof Sobkowiak JEE & OSS Architect | Technical Architect @ Capgemini Capgemini <http://www.pl.capgemini.com/> | Software Solutions Center <http://www.pl.capgemini-sdm.com/> | Wroclaw e-mail: [email protected] <mailto:[email protected]> | Twitter: @KSobkowiak Calendar: http://goo.gl/yvsebC
