[jira] [Resolved] (QPID-6201) BDB state change events must be delivered sequential to the virtualhostnode

["Alex Rudyy (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/QPID-6201?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alex Rudyy resolved QPID-6201.
------------------------------
    Resolution: Fixed

The changes look good

> BDB state change events must be delivered sequential to the virtualhostnode
> ---------------------------------------------------------------------------
>
>                 Key: QPID-6201
>                 URL: https://issues.apache.org/jira/browse/QPID-6201
>             Project: Qpid
>          Issue Type: Bug
>          Components: Java Broker
>    Affects Versions: 0.30
>            Reporter: Keith Wall
>            Assignee: Keith Wall
>              Labels: HA
>             Fix For: 0.31
>
>
> The ReplicatedEnvironmentFacade uses a thread pool executor to deliver JE state change notifications up to the BDBHAVHN.  JE requires this - it stipulates you should do minimal work in these callbacks.  These events (transitions to replica, master etc) need to be delivered sequentially to the node in order for the BDBHAVHN to behave correctly.
> Unfortunately a coding error has accidentally assigned this to a thread pool with multiple threads, meaning this may not be the case.
> This has been seen to cause a group in a state where a VHN was in the MASTER role, but its underlying virtual host remained UNAVAILABLE.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org