[
https://issues.jboss.org/browse/JGRP-1700?page=com.atlassian.jira.plugin....
]
Bela Ban commented on JGRP-1700:
--------------------------------
On resetDigest(), we don't use our own digest to see the digest, but leave all seqnos
as -1. When a STABLE message is received (also our own), we ignore the seqno if it is -1,
otherwise we do a min() for highest_delivered and a max() for highest_received.
This leads to more up-to-date STABILITY messages and thus faster purging of messages seen
by everyone.
STABLE: use a bit set instead of a hash set for votes
-----------------------------------------------------
Key: JGRP-1700
URL:
https://issues.jboss.org/browse/JGRP-1700
Project: JGroups
Issue Type: Enhancement
Reporter: Bela Ban
Assignee: Bela Ban
Fix For: 3.4
Currently, the votes hashset is cleared when the digest is reset, and then votes are
added as STABLE messages come in. However, since the digest is based on an Address[]
membership, we could create a FixedSizeBitSet, and the bit to set when a STABLE msg is
received is computed as the sender's rank in that membership. When all bits are set,
we can send a STABILITY message (= all votes have been received).
This takes up less space than a hashset.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:
http://www.atlassian.com/software/jira