AEM 6.2 Instance Hangs completely when packages with osgi config nodes are deployed along with Java Bundles.

[tarkash]

Hi,

We are using the AEM 6.2 and one issue that is observed is that we need to do a full restart of the AEM (kill java process)instance whenever we install packages which have config nodes. The instance gets totally hung and even restart of all bundles via "/system/console/vmstat: does not help. 

Also this issue is mostly observed on Publish instance and not that often on author. Though this is more of an observation only.

Just to add we have the following sling configurations that we are overriding in our package via osgi config nodes:

1. /apps/iea/config/org.apache.sling.jcr.resource.internal.JcrResourceResolverFactoryImpl

2. /apps/iea/config/org.apache.sling.serviceusermapping.impl.ServiceUserMapperImpl

3. /apps/iea/config/org.apache.sling.servlets.resolver.SlingServletResolver

[GK-AEM]

What's the start command are you using??

Are you starting the aem instance with enough heap memory size??

Thanks,

Kishore

[jörg_hoh1]

Hi,

Normally the system can sort it out by itself, what services have to be restarted in case you reconfigure services. In your case it obviously doesn't work.

Most often this has to do with references (most often custom code to other services), which are not handled properly (most often due to "caching"). Another problem could be that services have not been designed to be restarted, but most often just assume starting and stopping of the complete application. In that case the cleanup is not happending properly, which doesn't matter if the system is shutdown.

So I would start with the log files are carefully review what kind of exception you get and when the breakage begins. In parallel you should validate, that all your custom serivces and components can be restarted properly.

Jörg