[wildfly-dev] Completion of DeploymentUnitProcessor.deploy(ear, war1, war2) observed in a subsequent phase?
Peter Palaga
ppalaga at redhat.com
Wed Sep 19 03:14:50 EDT 2018
Very helpful, thanks a lot Brian and Stuart! -- P
On 18/09/18 23:41, Stuart Douglas wrote:
> For ear deployments each phase is first run for the ear, then run for
> all sub deployments in parallel. This means that any changes made by the
> ear deployer are visible, however you need to be careful about thread
> safety for the sub deployments if you are working with things attached
> to the top level ear, as multiple threads may be doing the same thing.
>
> Stuart
>
> On Wed, Sep 19, 2018 at 2:19 AM Peter Palaga <ppalaga at redhat.com
> <mailto:ppalaga at redhat.com>> wrote:
>
> Hi *,
>
> I cannot find any info about what (if any) are the guarantees for
> ordering and exit of DeploymentUnitProcessors (DUPs) in case of an EAR
> that contains multiple WARs.
>
> I have a situation like this: MyDUP1 is registered for Phase.PARSE and
> MyDup2 is registered for Phase.DEPENDENCIES.
> I can see from the logs that MyDUP1.deploy(ear),
> MyDUP1.deploy(ear.war1), MyDUP1.deploy(war2), etc. are executed by
> distinct threads which is perfectly OK.
> I'd like to know if there is any guarantee that MyDup2.deploy(*)
> observes the completion of MyDUP1.deploy(ear), MyDUP1.deploy(ear.war1),
> MyDUP1.deploy(ear.war2), etc.?
>
> MyDUP1.deploy() collects some info from the EAR and the WARs and stores
> it as an attachment of the DeploymentUnit. MyDup2.deploy() is supposed
> to read that attachment and so it is important that the info is
> complete
> when MyDup2.deploy() reads it.
>
> Thanks,
>
> -- Peter
> _______________________________________________
> wildfly-dev mailing list
> wildfly-dev at lists.jboss.org <mailto:wildfly-dev at lists.jboss.org>
> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>
More information about the wildfly-dev
mailing list