RpcDispatcher hangs waiting for response from dead members
----------------------------------------------------------
Key: JGRP-44
URL: http://jira.jboss.com/jira/browse/JGRP-44
Project: JGroups
Type: Bug
Versions: 2.2.8
Environment: Linux
Reporter: Mark Riley
Assigned to: Bela Ban
I am running four jvms with jgroups 2.2.1 across 2 Linux machines,
lets call them A & B on machine 1, and C & D on machine 2. They
all form into a single group. However if A & B are killed and
immediately afterwards C issues a remote method call (via
RpcDispatcher) to all members (i.e. before it receives the new
view), then C hangs, waiting for all responses, presumably from
the dead members A & B.
I removed FD_SOCK and inserted FD instead in the protocol stack and this
problem
seemed to go away. After doing this both A and B were marked as
suspected in the RPC. With FD_SOCK, RpcDispatcher never seemed to
receive the new view. It was as if the view with the dead members
got discarded somewhere along the line.
Just to clarify, FD successfully suspects the members every time,
FD_SOCK regularly fails to do so. They are being used in conjunction
with UDP.
FD properties:
"timeout" value="2000"
"max_tries" value="3"
"shun" value="true"
FD_SOCK has no properties set
--
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
-
If you want more information on JIRA, or have a bug to report see:
http://www.atlassian.com/software/jira
-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
JBoss-Development mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jboss-development
