[
https://jira.jboss.org/jira/browse/JGRP-853?page=com.atlassian.jira.plugi...
]
Bela Ban commented on JGRP-853:
-------------------------------
The reason is that we take the membership as the basis of picking the new ping
destination. This is wrong because some members of the membership might already be
suspected. In that case, we might return with null as ping_dest, and the pinger thread
terminates.
Solution: use a set that starts out as a copy of the membership, but remove all suspected
members.
Failure detection: multiple crashes not detected
------------------------------------------------
Key: JGRP-853
URL:
https://jira.jboss.org/jira/browse/JGRP-853
Project: JGroups
Issue Type: Bug
Reporter: Bela Ban
Assignee: Bela Ban
Fix For: 2.6.6
When many nodes in a cluster crash simultaneously, the surviving node doesn't change
its view.
To reproduce:
- Start 10 instances of Draw in a shell
- Start 1 instance in a different shell
- Kill shell #1
==> The view of the last instance doesn't go back to 1
This works fine in CVS head (2.7).
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira