[jboss-jira] [JBoss JIRA] Resolved: (JGRP-227) Distributed Hashtable call to put() blocks indefinitely

[Bela Ban (JIRA)]

     [ http://jira.jboss.com/jira/browse/JGRP-227?page=all ]

Bela Ban resolved JGRP-227.
---------------------------

    Resolution: Done

I ran the unit test 5 times, all 5 passed.

Okay, actually, I don't support calling put()/remove()/clear() etc from within a callback. As you mentioned, this has to be done on a separate thread, otherwise you will get into deadlocks, as described in http://www.jgroups.org/javagroupsnew/docs/papers/Middleware.ps.
So, while this is unsupported, I added a method setDeadlockDetection() to DistributedHashtable, which you can call with a "true" argument, before calling start(). That might help, let me know.

> Distributed Hashtable call to put() blocks indefinitely
> -------------------------------------------------------
>
>                 Key: JGRP-227
>                 URL: http://jira.jboss.com/jira/browse/JGRP-227
>             Project: JGroups
>          Issue Type: Bug
>    Affects Versions: 2.2.9.1
>         Environment: Mandrake Linux 9.1,  Java 1.4.2
>            Reporter: Mark Kopec
>         Assigned To: Bela Ban
>             Fix For: 2.4
>
>         Attachments: JGroupsDistTest.java
>
>
> Distributed hashtable put(x,y) without any ERROR level locks up and creates inconsistent DHT values across cluster.
> Here is the code that exposes this error:
> MyListener implements DistributedHashtable.notification{
> // other methods
>     public void entryRemoved(Object key) {
>         distributedHashTableObject.put(key,"xxxsomeval"); //<-- this and subsequent on the distributedHashTableObject put() or remove() get stuck!!!!
>  }

-- 
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