Re: [pax-cdi] Thread dead lock with osgi refresh

I pushed a fix anyway. 2017-05-10 13:50 GMT+02:00 Pavel : > Simon said which configuration must be used. Weld developers said this > configuration > can be set via system settings -Dorg.jboss.weld.bootstrap. > concurrentDeployment=false > so there is no need to modify

Re: [pax-cdi] Thread dead lock with osgi refresh

Simon said which configuration must be used. Weld developers said this configuration can be set via system settings -Dorg.jboss.weld.bootstrap.concurrentDeployment=false so there is no need to modify PAX-CDI code. If someone has the same problem he can find more information here:

Re: [pax-cdi] Thread dead lock with osgi refresh

Hi Simon Thank you very much! This is really helped. But what are the consequences of such workaround? Best regards, Pavel вторник, 9 мая 2017 г., 18:43:43 UTC+3 пользователь Simon Spero написал: > > This is probably suboptimal, but it may be good enough: > > When configuring the weld

Re: [pax-cdi] Thread dead lock with osgi refresh

This is probably suboptimal, but it may be good enough: When configuring the weld bootstrap at https://github.com/ops4j/org.ops4j.pax.cdi/blob/master/pax-cdi-weld/src/main/java/org/ops4j/pax/cdi/weld/impl/WeldCdiContainer.java#L120 it is possible to disable concurrent deployment by chaining

Re: [pax-cdi] Thread dead lock with osgi refresh

Guillaume, thank you for your suggestions, but it didn't help. I've checked all bundles manifest and only in two I found dynamic import in felix.scr and org.osgi.enterprise. I removed those dynamic imports but it didn't help. Result is the same. вторник, 9 мая 2017 г., 16:20:49 UTC+3

Re: [pax-cdi] Thread dead lock with osgi refresh

Or could you put a breakpoint where I told you, that would give you the name of the package and the bundle by looking at the variables on the stack. 2017-05-08 12:02 GMT+02:00 Pavel : > Hi all > > I use pax-cdi 1.0.0.RC2 with weld-osgi-bundle 2.3.5. And I have two >

Re: [pax-cdi] Thread dead lock with osgi refresh

As I understand you DynamicImport-Package must be defined in bundle MANIFEST. So do I need to check MANIFESTs of all bundles for containing DynamicImport-Package property? вторник, 9 мая 2017 г., 16:04:53 UTC+3 пользователь Guillaume Nodet написал: > > Well, according to the stack trace,

Re: [pax-cdi] Thread dead lock with osgi refresh

Well, according to the stack trace, someone does. I've checked weld and pax-cdi bundles and they don't have any dynamic import. Can you double check all the deployed bundles ? Maybe it's not one of you bundles, but there's definitely at least one bundle which uses dynamic import. One way to

Re: [pax-cdi] Thread dead lock with osgi refresh

Hi Guillaume Thank you for your answer. But I don't use any dynamic-imports. вторник, 9 мая 2017 г., 15:52:03 UTC+3 пользователь Guillaume Nodet написал: > > Yeah, and I'm telling you that you should remove the DynamicImport from > your bundles in order maybe to get around those issues. > >

Re: [pax-cdi] Thread dead lock with osgi refresh

Yeah, and I'm telling you that you should remove the DynamicImport from your bundles in order maybe to get around those issues. 2017-05-09 14:40 GMT+02:00 Pavel : > I wrote to Weld developers and this is the answer I got from them > > I'm not sure whether this is a

Re: [pax-cdi] Thread dead lock with osgi refresh

I wrote to Weld developers and this is the answer I got from them I'm not sure whether this is a PAX-CDI, Felix or Weld problem. The only argument is what I wrote earlier: "it is true that Weld's ClassFileUtils holds 0xfd03aa50 and also waiting in another thread to release this lock. But

Re: [pax-cdi] Thread dead lock with osgi refresh

You should avoid the use of dynamic imports whenever possible. Try removing those from your bundles. 2017-05-08 12:02 GMT+02:00 Pavel : > Hi all > > I use pax-cdi 1.0.0.RC2 with weld-osgi-bundle 2.3.5. And I have two > bundles: bundleA and bundleB. > > BundleB has CDI

Re: [pax-cdi] Thread dead lock with osgi refresh

I see two locks in JBoss Weld. - locked <0xfd03aa50> (a java.lang.Class for org.jboss.weld.util.bytecode.ClassFileUtils) and I doubt that it is anything we can do about it in OPS4J to avoid this. On Mon, May 8, 2017 at 6:02 PM, Pavel wrote: > Hi all > > I