[
https://issues.apache.org/jira/browse/CASSANDRA-15349?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16968504#comment-16968504
]
Chris Lohfink commented on CASSANDRA-15349:
-------------------------------------------
I kinda disagree with 3. I think if the client _can_ disconnect it can result
in a more responsive shutdown than magic 10 second sleeps. More of a
1. stops accepting new connections
2. sends {{GOING_AWAY}}
3. waits for either there are no more connections or a {{max(readtimeout,
writetimeout)}} timeout
4. close remaining connections and continue
also between 2 and 4 could we return server overloaded or something to make it
retry on different hosts incase queries do come in? Maybe make shutdown more
clean for versions of client previous to ones that support this feature.
> Add “Going away” message to the client protocol
> -----------------------------------------------
>
> Key: CASSANDRA-15349
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15349
> Project: Cassandra
> Issue Type: New Feature
> Components: Messaging/Client
> Reporter: Alex Petrov
> Priority: Normal
> Labels: client-impacting
>
> Add “Going away” message that allows node to announce its shutdown and let
> clients gracefully shutdown and not attempt further requests.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
