[
https://issues.apache.org/jira/browse/CASSANDRA-13713?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16111375#comment-16111375
]
Joel Knighton edited comment on CASSANDRA-13713 at 8/2/17 5:31 PM:
-------------------------------------------------------------------
Ah - guess I missed that one. LGTM to the change here. I don't have any strong
opinions as to where it goes in 2.2+. While we don't have any known problems
related to this, we also know bugs can lurk in this subsystem for a long time,
and this seems like a safe preventative measure.
was (Author: jkni):
Ah - guess I missed that one. +1 to the change here. I don't have any strong
opinions as to where it goes in 2.2+. While we don't have any known problems
related to this, we also know bugs can lurk in this subsystem for a long time,
and this seems like a safe preventative measure.
> Move processing of EchoMessage response to gossip stage
> -------------------------------------------------------
>
> Key: CASSANDRA-13713
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13713
> Project: Cassandra
> Issue Type: Bug
> Components: Distributed Metadata
> Reporter: Jason Brown
> Assignee: Jason Brown
> Priority: Minor
> Fix For: 4.x
>
>
> Currently, when a node receives an {{EchoMessage}}, is sends a simple ACK
> reply back (see {{EchoVerbHandler}}). The ACK is sent on the small message
> connection, and because it is 'generically' typed as
> {{Verb.REQUEST_RESPONSE}}, is consumed on a {{Stage.REQUEST_RESPONSE}}
> thread. The proper thread for this response to be consumed is
> {{Stage.GOSSIP}}, that way we can move more of the updating of the gossip
> state to a single, centralized thread, and less abuse of gossip's shared
> mutable state can occur.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
