TP: fetch physical address associated with logical address more quickly
-----------------------------------------------------------------------
Key: JGRP-1012
URL:
https://jira.jboss.org/jira/browse/JGRP-1012
Project: JGroups
Issue Type: Task
Reporter: Bela Ban
Assignee: Bela Ban
Fix For: 2.8
Currently, when we send a unicast message and don't have the physical address
available, we send a discovery request for the physical address and trash the message.
Later, when the physical address has been received and added to the transport's cache,
the retransmission of the unicast message will succeed.
In order not to generate too much traffic, we send such a discovery request only every 5
seconds (hard-coded) max. However, this is for *all* addresses. So if we have physical
addresses for P, Q and R missing, the total time to send the 3 discovery requests will be
15 sec max ! So we would not be able to send a unicast to R for up to 15 secs !
This is bad, we should keep track of *which* addresses we've sent out discovery
requests for, and have timeouts based on individual addresses, for example in the form of
a cache (keys=addrs, values=last time sent).
--
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