[jboss-jira] [JBoss JIRA] (WFCORE-2331) When --cached-dc used HC does not inform on configuration change if DC reconnection happens

[Ken Wills (JIRA)]

     [ https://issues.jboss.org/browse/WFCORE-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ken Wills moved JBEAP-9133 to WFCORE-2331:
------------------------------------------

              Project: WildFly Core  (was: JBoss Enterprise Application Platform)
                  Key: WFCORE-2331  (was: JBEAP-9133)
             Workflow: GIT Pull Request workflow   (was: CDW with loose statuses v1)
          Component/s: Domain Management
                           (was: Domain Management)
    Affects Version/s:     (was: 7.1.0.DR11)


> When --cached-dc used HC does not inform on configuration change if DC reconnection happens
> -------------------------------------------------------------------------------------------
>
>                 Key: WFCORE-2331
>                 URL: https://issues.jboss.org/browse/WFCORE-2331
>             Project: WildFly Core
>          Issue Type: Bug
>          Components: Domain Management
>            Reporter: Ken Wills
>            Assignee: Ken Wills
>
> HC does not inform that DC changed configuration and reload/restart is required when `--cached-dc` is used.
> This issue was created based on discussion at https://issues.jboss.org/browse/EAP7-496.
> HC starts with `--cached-dc` and DC is down. If DC contains changed attribute and is started HC receives configuration updates that was changed in comparison to existing 
> "domain.cached-remote.xml". The change seems to be propagated to HC but it's not "activated" (it's not part of model at HC) neither information that reload/restart of HC is needed is provided.
> This could be checked with following steps
> # start DC and HC with {{--cached-dc}} parameter
>  {code}
>  cd $JBOSS_HOME
>  ./bin/domain.sh --host-config=host-master.xml
>  ./bin/domain.sh  --host-config=host-slave.xml -Djboss.domain.master.address=127.0.0.1 -Djboss.management.native.port=9899 -Djboss.host.name=slave --cached-dc
>  {code}
> # stop HC
> # configure logging at DC
>  {code}
>  cd $JBOSS_HOME
>  ./bin/jboss-cli.sh -c
>  /profile=full/subsystem=logging/logger=my.test:add(category=my.test, level=FINE)
>  {code}
> # stop DC
> # start HC
>  {code}
>  ./bin/domain.sh  --host-config=host-slave.xml -Djboss.domain.master.address=127.0.0.1 -Djboss.management.native.port=9899 -Djboss.host.name=slave --cached-dc
>  {code}
> # check that logging is not configured
>  {code}
>  ./bin/jboss-cli.sh -c --controller=remoting://localhost:9899
>  /host=slave/server=server-one/subsystem=logging:read-children-resources(child-type=logger)
>  {code}
> # start DC and wait a while when HC is connected to it
>  {code}
>  INFO  [org.jboss.as.host.controller] (Host Controller Service Threads - 33) WFLYHC0150: Trying to reconnect to master host controller.
>  INFO  [org.jboss.as.host.controller] (Host Controller Service Threads - 33) WFLYHC0148: Connected to master host controller at remote://127.0.0.1:9999
>  {code}
> # check what DC says about server-one
>  {code}
>  ./bin/jboss-cli.sh -c
>  /host=slave/server=server-one/subsystem=logging:read-children-resources(child-type=logger)
>  {code}
> The result is that the new logger {{my.test}} is not part of the configuration of {{server-one}}. There is no warning about such fact in server log what I can see.
> True is that here behaves differently against my testing before that after printing resources of the subsystem {{logging}} node there is information {{reload-required}}.
> From your point of view - is this correct behavior? If so I think some warning message in server log at least that content of {{domain.cached-remote.xml}} is not in sync with servers' model should be promoted.



--
This message was sent by Atlassian JIRA
(v7.2.3#72005)