[
https://issues.apache.org/jira/browse/CASSANDRA-2961?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13083281#comment-13083281
]
Brandon Williams commented on CASSANDRA-2961:
---------------------------------------------
It looks like as a benefit of our ghetto string concatenation/delimiting in
VersionedValue, we can just append timestamps to states that need to expire,
then respect those in the gossiper (if present) instead of aVLT, but we still
need to use aVLT if not. This way, everything will stay backward-compatible,
since older nodes will just get extra 'pieces' in SS notifications and not use
them. We probably only need to do this for the removed and left states, since
the fat client removal logic is already solid and a timestamp won't really help
there anyway.
> Expire dead gossip states based on time
> ---------------------------------------
>
> Key: CASSANDRA-2961
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2961
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Brandon Williams
> Assignee: Patricio Echague
>
> Currently dead states are held until aVeryLongTime, 3 days. The problem is
> that if a node reboots within this period, it begins a new 3 days and will
> repopulate the ring with the dead state. While mostly harmless, perpetuating
> the state forever is at least wasting a small amount of bandwidth. Instead,
> we can expire states based on a ttl, which will require that the cluster be
> loosely time synced; within the quarantine period of 60s.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
