Hi,
We recently migrated our platform from Karaf 2.3.2 to Karaf 3.0.3.
Everything runs fine excepted JPA and Hibernate.
When we start our custom distribution in some environments, persistence
bundles stay in GracePeriod and are never initialized, as if the persistence
context was never read. They end in the state "Failure".
It's a strange comportment because it's not happen on the same rate,
depending on the environment where the distribution is launched. On my
development machine, the distribution runs almost always fine. On one of our
staging environments, almost never ...
I noticed some suspect logs :
2015-05-18 14:42:47,653 | WARN | Event Dispatcher | aries
| 192 - org.apache.aries.jpa.blueprint.aries - 1.0.4 | Managed persistence
context support is no longer available for use with the Aries Blueprint
container.
2015-05-18 14:42:48,037 | WARN | Event Dispatcher | container
| 193 - org.apache.aries.jpa.container - 1.0.2 | There are no providers
available.
2015-05-18 14:43:12,443 | ERROR | Event Dispatcher | aries
| 192 - org.apache.aries.jpa.blueprint.aries - 1.0.4 | No
org.apache.aries.jpa.container.sync.Synchronization service registered.
But these logs appear also when distribution run fine, so maybe it's not
related to the problem.
I also noticed that in case of "GracePeriod then Failure" state, the bundle
"hibernate-osgi" stays indefinitely in the state "Starting".
And the bundle "Apache Aries Blueprint Core Compatiblity Fragment Bundle,
Hosts: 15" stays in the state Resolved. I don't know what is its purpose and
if it could be linked to our problem.
Here an example of one our persistence features :
<feature name="usersDAO" version="${project.version}" description="usersDAO
feature">
<feature>transaction</feature>
<feature>jndi</feature>
<feature>pax-jdbc-h2</feature>
<feature>pax-jdbc-config</feature>
<feature>pax-jdbc-pool-dbcp2</feature>
<feature version="2.1.0">jpa</feature>
<feature version="4.3.6.Final">hibernate</feature>
<bundle>mvn:com.test.bundle.dao/users/1.1.2</bundle>
</feature>
As you can see it's very close to this example :
https://github.com/cschneider/Karaf-Tutorial/tree/master/db/examplejpa
So, do you have an idea ?
I suspect the bundle order, or the start-level of some bundles. Maybe the
distribution runs too fast, or too slow, on some environments, and so
initialize some importants bundles too fast, or too slowly ?
Thank you for your help.
--
View this message in context:
http://karaf.922171.n3.nabble.com/Karaf-3-0-3-Hibernate-JPA-JPA-bundles-stay-in-GracePeriod-state-tp4040421.html
Sent from the Karaf - User mailing list archive at Nabble.com.
