[ 
https://issues.apache.org/jira/browse/CASSANDRA-16880?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Josh McKenzie updated CASSANDRA-16880:
--------------------------------------
    Test and Documentation Plan: No new testing nor documentation needed; 
trivial change.
                         Status: Patch Available  (was: Open)

2 test failures:
 # JDK11_unit: testGetPositionsKeyCacheStats, which passes locally for me in 
IDE + cmd line and appears unrelated to this diff
 # JDK11 dtest no vnode, with a teardown error on 
test_complementary_deletion_with_limit_on_static_column_with_empty_partitions 
which I also cannot reproduce locally and appears unrelated to this diff.

Pending discussion on the ML about trivial improvements we'll decide where to 
merge this (4.0.x vs. 4.x); should be clean diff to either.
||Item|Link|
|JDK8 
tests|[Link|https://app.circleci.com/pipelines/github/josh-mckenzie/cassandra/60/workflows/67473fbc-88f7-44d3-a409-b616e2cadbb4]|
|JDK11 
tests|[Link|https://app.circleci.com/pipelines/github/josh-mckenzie/cassandra/60/workflows/15e09ea4-4d35-4035-9afc-ff0d1089041e]|
|Branch|[Link|https://github.com/apache/cassandra/compare/cassandra-4.0...josh-mckenzie:CASSANDRA-16880?expand=1]|

> Catch read repair timeouts and add metrics to indicate they occurred
> --------------------------------------------------------------------
>
>                 Key: CASSANDRA-16880
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-16880
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Observability/Metrics
>            Reporter: Josh McKenzie
>            Assignee: Josh McKenzie
>            Priority: Normal
>             Fix For: 4.0.x
>
>
> When we fire off async read repairs onto their own executor they may time out 
> and in doing so, we don't have anything that stops them from propagating that 
> timeout exception the way up to CassandraDaemon's uncaught exception handler. 
> When this happens we logs at ERROR.
> Obviously a timeout isn't great, but it's not an ERROR, so we should trap 
> them instead and add some metrics around this occurrance.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to