cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ariel Weisberg (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-9237) Gossip messages subject to head of line blocking by other intra-cluster traffic
Date Sat, 25 Apr 2015 15:06:38 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-9237?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14512526#comment-14512526
] 

Ariel Weisberg commented on CASSANDRA-9237:
-------------------------------------------

That will only address some of the head of line blocking scenarios. Lot's of small messages
on a low throughput connection will can still block gossip. Now if a connection is that low
throughput does it matter that gossip messages got through? I don't know.

Right now I am leaning towards allowing all received network messages from a cluster node
to function as indication of liveness and reachability for failure detector. This is something
that is easy to implement since it just means walking all the connections and checking the
last received timestamp.

At that point Gossip heartbeats would just be for gossip and as a backup when the cluster
is idle.

> 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)

Mime
View raw message