[jboss-jira] [JBoss JIRA] Commented: (JBCACHE-1165) Endless loop in PessimisticLockInterceptor, sometimes seen as a NodeNotExistsException

[Manik Surtani (JIRA)]

    [ http://jira.jboss.com/jira/browse/JBCACHE-1165?page=comments#action_12387719 ] 
            
Manik Surtani commented on JBCACHE-1165:
----------------------------------------

The fix is not absolute and will require pretty bug architectural changes.

There is an edge case where a NodeNotFoundException may occur, and this is due to parent nodes not being
write locked when children are added/removed.

The problem is in the way READ_COMMITTED is implemented, i.e., writers are not blocked by readers and this
allows a reader to hold a lock when a writer comes in and deletes the node in question.

Workaround is to use LockParentForChildInsertRemove OR to use REPEATABLE_READ.



> Endless loop in PessimisticLockInterceptor, sometimes seen as a NodeNotExistsException
> --------------------------------------------------------------------------------------
>
>                 Key: JBCACHE-1165
>                 URL: http://jira.jboss.com/jira/browse/JBCACHE-1165
>             Project: JBoss Cache
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>    Affects Versions: 2.0.0.GA, 1.4.1.SP4
>         Environment: Windows XP, Sun jdk150_06 
> Solaris 10
>            Reporter: Jacek Halat
>         Assigned To: Manik Surtani
>             Fix For: 2.1.0.GA, 1.4.1.SP7
>
>         Attachments: ConcurentPutRemoveTest.java, ConsoleOut.log
>
>
> See ConcurrentPutRemoveTest
> This sometimes manifests itself as an endless loop, and sometimes as a NodeNotExistsException (on the putter thread)  when concurrent puts and removes are involved.

-- 
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