[
https://issues.jboss.org/browse/JGRP-1403?page=com.atlassian.jira.plugin....
]
Witold Wojciechowski commented on JGRP-1403:
--------------------------------------------
Bela, thanks for looking into this. Please keep in mind that we are running on an older
version (2.8), which has been working fine for us for 2 years. We will consider moving to
newer one, if it makes a difference, but are hesitant to do so as a matter of routine
because of the ever present risk if destabilizing a system which behaves well in all other
respects. This issue was discovered only recently during a process of network access
audit. Our networking group tries to keep things pristine, and does not want to see minor
but unexpected anomalies.
JGroups diagnostics opens listening sockets on all interfaces on a
multi-homed host
-----------------------------------------------------------------------------------
Key: JGRP-1403
URL:
https://issues.jboss.org/browse/JGRP-1403
Project: JGroups
Issue Type: Bug
Affects Versions: 2.8
Environment: Linux
Reporter: Witold Wojciechowski
Assignee: Bela Ban
Fix For: 3.1
Netstat Linux utility reports that my application using JGroups opens sockets on all
multicast-capable interfaces and subscribes to a multicast address specified for use by
diagnostics (per documentation), despite explicit specification of the bind_addr property
as Java system property. That is undocumented, and unexpected. (The setting of bind_addr
has the expected effect for the regular, JGroups data sockets.)
We tried to disable diagnostics altogether, by using the enable_diagnostics=false, but it
makes no difference.
We tried to set receive_on_all_interfaces=false, but it makes no difference,
We also tried to set the the receive_interfaces=<our desired interface IP address>
besides setting bind_addr, and it also makes no difference.
Our application does not use the xml configuration file for the JGroups protocol stack,
and we do it programmatically via setting appropriate Java system properties instead,
before any JGroups functionality is invoked. This is done so, because the multicast
addresses our application uses are not static (are determined at run time).
I am not attaching any sample code, since any sample program when run on a multi-homed
machine will reproduce the problem.
Any comments on this behaviour would be greatly appreciated.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see:
http://www.atlassian.com/software/jira