[jira] [Commented] (CASSANDRA-13124) Send ack when received hint is non-local

["Paulo Motta (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-13124?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15823340#comment-15823340 ] 

Paulo Motta commented on CASSANDRA-13124:
-----------------------------------------

continue discussion from CASSANDRA-13058 (cc [~spodxx@gmail.com]):

bq. and terminate with a successful return value afterwards

Does this mean retry is broken on 3.0? Should we file a ticket for this? From what I understood CASSANDRA-11960 only affected 3.X so I guess retry should be working on 3.0?

If my understanding above is correct we should probably also fix this on 3.0 to reinstate the same behavior before CASSANDRA-12905? Can you create a 3.0 patch and submit CI?

bq. That being said, it's less than ideal that dtests annotated with "@no_vnodes" are simply skipped during ad-hoc job scheduling during reviews.

Agreed, created the following [issue|https://github.com/riptano/cassandra-dtest/issues/1422] on cassandra-dtest to address this, feel welcome to give feedback there.

> Send ack when received hint is non-local
> ----------------------------------------
>
>                 Key: CASSANDRA-13124
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-13124
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Paulo Motta
>            Priority: Minor
>
> CASSANDRA-12905 changed the hints path to be asynchronous in the normal case, but when the hint is non-local and should be stored (ie. on decommission) an ack is not sent so the operation does not complete successfully.
> CASSANDRA-13058 fixed this for 3.0+, but for 3.11+ but there is an additional concern which on 3.0 is that non-acked hints are being completed successfully so we should probably look into this as well here.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)