[jboss-jira] [JBoss JIRA] Updated: (JBCACHE-1209) Restore existing Option to InvocationContext in TreeCache._gravitateData

[Brian Stansberry (JIRA)]

     [ http://jira.jboss.com/jira/browse/JBCACHE-1209?page=all ]

Brian Stansberry updated JBCACHE-1209:
--------------------------------------

    Attachment: TreeCache.patch

Possible patch.  With this in place, the 150 AS testsuite failures I was seeing went away.

Running the full JBC testsuite w/ this now (w/ 1.4.1.SP6 tag).

The above patch may be a bit of a sledgehammer; perhaps just the stuff in the finally block would take care of the problem. But I figured restoring the Option after each get(Fqn) call was just safer.

> Restore existing Option to InvocationContext in TreeCache._gravitateData
> ------------------------------------------------------------------------
>
>                 Key: JBCACHE-1209
>                 URL: http://jira.jboss.com/jira/browse/JBCACHE-1209
>             Project: JBoss Cache
>          Issue Type: Sub-task
>      Security Level: Public(Everyone can see) 
>    Affects Versions: 1.4.1.SP6
>            Reporter: Brian Stansberry
>         Assigned To: Manik Surtani
>             Fix For: 1.4.1.SP6
>
>         Attachments: TreeCache.patch
>
>
> The fix for JBCACHE-1192 has introduced a couple of problems that are manifesting themselves in numerous errors in the test suite:
> 1) _gravitateData changes the value of skipDataGravitation for the Option associated with the call, but then doesn't restore the correct value to *that* instance of Option.
> 2) _gravitateData calls get(Fqn), and that call actually results in the instance of Option that is associated with the thread being changed (see impl of get(Fqn). But, _gravitateData does not restore the original Option to the thread.  Maybe not a problem if _gravitateData is called remotely, since there is no meaningful Option associated with the thread.  But _gravitateData can now also be called by DataGravitationInterceptor.localBackupGet, and in that case there very well could be a meaningful Option associated with the thread, and it needs to be restored before _gravitateData returns.
> I'll attach a possible patch in a minute.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira