[jboss-jira] [JBoss JIRA] (WFCORE-297) HC should remember the 'run' state of the server instances after crash or shutdown

[Brian Stansberry (JIRA)]

    [ https://issues.jboss.org/browse/WFCORE-297?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13032626#comment-13032626 ] 

Brian Stansberry commented on WFCORE-297:
-----------------------------------------

Emmanuel, we may be able to work around this using a custom Resource impl for the server-config resource. Pass the ServerInventory into it via it's constructor. If ''set-auto-start-on-start-stop' is true, the Resource.getModel() method will determine the server status from the inventory and use it to set the value of 'auto-start' before returning the model.

This will need care though, as getting the right value from the server inventory is not as simple as it may seem. For example, this in the CLI could produce a wrong result:

{code}
batch
/host=x/system-property=y:add(value=1)
shutdown --host=x
run-batch
{code}

The system property add (or any config change to host=*) will create a need for persisting the xml. But by the time the composite op gets to point where it reads the model to persist to xml, the stop of the servers may have begun, and this could result in auto-start=false getting incorrectly written.

Basically, this would need to be able to discriminate server stops triggered by an admin's server=xxx:stop ops from those caused by the HC itself stopping.

> HC should remember the 'run' state of the server instances after crash or shutdown
> ----------------------------------------------------------------------------------
>
>                 Key: WFCORE-297
>                 URL: https://issues.jboss.org/browse/WFCORE-297
>             Project: WildFly Core
>          Issue Type: Feature Request
>          Components: Domain Management
>            Reporter: Wolf-Dieter Fink
>            Assignee: Emmanuel Hugonnet
>              Labels: EAP, todo
>
> The host controller should save which server is currently up and running. This would allow the host controller to bring up all previously running instances on a restart.
> The idea is to support the same behavior that other application server (i.e WebLogic) supports.
> If a server is started or stopped during the lifetime of the DC/HC it should be in the same state after shutdown the DC/HC or a system crash.
> This can be achieved by an optional flag 'set-auto-start-on-start-stop' where the default is false which is the current behaviour.
> If set to true, a start of the server instance will set auto-start=true and a stop auto-start=false.
> If the server should not be started after a crash for any reason, this can be simple done by setting auto-start=false within the configuration, after starting the server the flag will be set b/c of the 'set-auto-start-on-start-stop' flag.



--
This message was sent by Atlassian JIRA
(v6.3.11#6341)