[
https://issues.apache.org/jira/browse/CASSANDRA-20505?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Josh McKenzie updated CASSANDRA-20505:
--------------------------------------
Test and Documentation Plan: Extensive property testing of callbacks
Status: Patch Available (was: Open)
> Log replicas on timeout; property test RequestCallback implementers
> -------------------------------------------------------------------
>
> Key: CASSANDRA-20505
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20505
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Messaging/Internode
> Reporter: Josh McKenzie
> Assignee: Josh McKenzie
> Priority: Normal
> Fix For: 5.x, 6.x
>
>
> We inconsistently log which replicas are involved in timeouts in a couple of
> our callbacks and don't log details in the vast majority; this is something
> we should do for all timeouts or failures for purposes of troubleshooting in
> prod clusters.
> Factoring this code out into a shared class so we don't copy-paste duplicate
> throughout all our callbacks, which we've done historically, also
> necessitates moving more logic out of the callbacks into a shared place to
> prevent duplicate operations (counting of responses, success or failure
> determination, etc). This also provides an opportunity to start bending
> towards standardization in our callbacks on how we quantify success, how we
> keep track of per-DC responses, and a host of other small things that are
> duplicated across our internode callbacks.
> These callbacks don't have much unit testing outside the
> {{WriteResponseHandlerTest}} that's primarily dealing with Ideal
> ConsistencyLevel metric calculation, so we should also introduce robust
> property testing for our primary callbacks.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
