[
https://issues.apache.org/jira/browse/CASSANDRA-8592?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14353917#comment-14353917
]
Tyler Hobbs commented on CASSANDRA-8592:
----------------------------------------
Overall the changes look good so far. Here are some preliminary review
comments and nitpicks:
* HintedHandoffManager:
** Use a single log call for the error message and exception info where
possible to avoid splitting the two parts in the logs
* MutationVerbHandler:
** I can't see a good reason to wrap the exception in an IOError. It should be
fine to re-throw the exception (there's only one caller of this method).
* Commitlog:
** Unrelated whitespace change
* WriteFailureException
** The class-level curly brace should go on a new line
* AbstractReplicationStrategy.getWriteResponseHandler():
** There is trailing whitespace on the changed lines
* MessageDeliveryTask:
** The error message refers to mutations, but this code path also handles
reads. Plus, mutation errors should already be logged by MutationVerbHandle.
** The main reason for throwing non-TombstoneOverwhelmingExceptions is to log
the full error and stacktrace. Perhaps we should just keep this and remove the
above error logging and the try/catch error logging in MutationVerbHandler?
* AbstractWriteResponseHandler:
** There is a TODO about moving a new method to a utility class
* StorageProxy:
** In mutate(), mention the number of failures in the WriteFailureException
Tracing entry
** Some trailing whitespace in the javadoc for hintMutations. Also, it seems
like it would be more natural to keep the ConsistencyLevel.ANY check in the
exception handlers instead of in the hintMutations() method so that it's clear
why we do or do not need to record a failure.
** In insertLocal(), you're missing whitespace between "catch" and "(Exception"
> Add WriteFailureException
> -------------------------
>
> Key: CASSANDRA-8592
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8592
> Project: Cassandra
> Issue Type: New Feature
> Components: API, Core
> Reporter: Tyler Hobbs
> Assignee: Stefania
> Labels: client-impacting
> Fix For: 3.0
>
>
> Similar to what CASSANDRA-7886 did for reads, we should add a
> WriteFailureException and have replicas signal a failure while handling a
> write to the coordinator.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
