[
https://issues.jboss.org/browse/JGRP-2029?page=com.atlassian.jira.plugin....
]
Bela Ban commented on JGRP-2029:
--------------------------------
You have a somewhat unusual configuration, e.g.
* SEQUENCER is now where it is supposed to be (below FRAG2)
* FLUSH is in the config; this combo has never been tested. I suggest remove FLUSH
* PEER_LOCK: not recommened, CENTRAL_LOCK is the recommended lock protocol
SEQUENCER locks itself from receiving messages
----------------------------------------------
Key: JGRP-2029
URL:
https://issues.jboss.org/browse/JGRP-2029
Project: JGroups
Issue Type: Bug
Affects Versions: 3.6.1
Environment: two nodes cluster environment, we are using infinispan 7.1.1 with
JGroups 3.6.1-FINAL inside.
Reporter: Sean Guo
Assignee: Bela Ban
Fix For: 3.6.11, 4.0
Attachments: filtered_threads.txt, jgroups-tcp.xml
I attached the filtered thread dump and the jgroups-tcp.xml from one node.
If I read the code correctly, all the threads reading message from Socket are
blocked(INT-1...INT-4) and they are waiting the lock owned by #489 thread.
Please check whether this is a bug or configuration problem. Currently we have to remove
the SEQUENCER from JGroups' protocol stack.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)