[
https://issues.apache.org/jira/browse/TORQUE-163?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13060351#comment-13060351
]
Thomas Vandahl commented on TORQUE-163:
---------------------------------------
Actually, the code in TorqueInstance.java (line 164ff) was supposed to handle
(or rather work around) a similar case. Obviously, the removal of the cache
entry in line 691 breaks this intention. Torque has no means of knowing which
peers actually exist right now. Do we want to change that? It would probably
have lots of implications.
Bye, Thomas.
> Map builders are emptied on Torque.shutdown() and not rebuilt on subsequent
> init()
> ----------------------------------------------------------------------------------
>
> Key: TORQUE-163
> URL: https://issues.apache.org/jira/browse/TORQUE-163
> Project: Torque
> Issue Type: Bug
> Affects Versions: 3.3
> Reporter: Thomas Fox
> Assignee: Thomas Fox
> Fix For: 4.0
>
>
> Problem: After a shutdown() and init() of Torque, the Map Builder cache
> entries which have been present before shutdown are not present anymore after
> a new init.
> Analysis: If a Peer class is loaded, it registers its map builder and the
> MapBuilder is built immediately or on Torque initialisation, depending on
> whether Torque is initialized or not.
> On Torque.shutdown(), all known map builders are removed.
> On a new init(), these Map builders will not be rebuilt anew because the Peer
> class will not be loaded a second time.
> Solution: The Map Builder cache entries should not be removed on
> Torque.shutdown()
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
