[
https://issues.apache.org/jira/browse/CASSANDRA-9237?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14512605#comment-14512605
]
Benedict commented on CASSANDRA-9237:
-------------------------------------
That's a good point. But since one of the ramifications of this is to truncate
the outbound message queue and then attempt to reconnect, it's likely we only
make things much worse. So at least avoiding that would be a good thing. I
haven't thought about it too much, but I think Ariel's proposal of not closing
a connection to a host if we're still receiving messages from it (and we're not
blocked writing messages to it) makes sense. Whether or not we mark it as down
for other purposes is another question. If we cannot get messages through (or
vica-versa) then probably we do want to hint and avoid routing messages to it.
> Gossip messages subject to head of line blocking by other intra-cluster
> traffic
> -------------------------------------------------------------------------------
>
> Key: CASSANDRA-9237
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9237
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Ariel Weisberg
> Assignee: Ariel Weisberg
>
> Reported as an issue over less than perfect networks like VPNs between data
> centers.
> Gossip goes over the small message socket where small is 64k which isn't
> particularly small. This is done for performance to keep most traffic on one
> hot socket.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
