[
http://jira.jboss.com/jira/browse/JGRP-547?page=comments#action_12367984 ]
Bela Ban commented on JGRP-547:
-------------------------------
I changed some code in ConnectionTableNIO which closes existing connections instead of
reusing them. This might produce the issue you're seeing. This is in CVS head.
However, we will only really test TCP_NIO in 2.6
(
http://jira.jboss.com/jira/browse/JGRP-548 includes it in the cruisecontrol runs), so for
2.5 we recommend TCP. If you're concerned about scaling, in [1] we achieved excellent
results with TCP:MPING for 8 node clusters.
[1]
http://www.jgroups.org/javagroupsnew/perfnew/Report.html
ReplicatedHashMap does not synchonized state with the members (CSV
tips)
------------------------------------------------------------------------
Key: JGRP-547
URL:
http://jira.jboss.com/jira/browse/JGRP-547
Project: JGroups
Issue Type: Bug
Reporter: Dipak Kothari
Assigned To: Bela Ban
Fix For: 2.5
Attachments: loadtest.zip
As per the message trail in the JGroup Help forum, starting many services concurrently
causes the state not to synchronize - have noticed more then one cordinator. When the
merge happens the state is not synchronized or I don't know how to. I have attached
example code that causes this behaviour on my pc.
The JGroup jar contains source from 2.5cr1 release with the ReplicatedHashMap added in.
The loadtest package contains my code, pretty self explainatory
--
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