[ 
https://issues.apache.org/jira/browse/CASSANDRA-19046?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17807911#comment-17807911
 ] 

Brandon Williams commented on CASSANDRA-19046:
----------------------------------------------

I've been looking into the logging, but it hasn't been easy to locate where 
that 'Paxos Commit' string is being set in v1.

As for the metrics, 

bq.  I'm honestly not 100% sure it's a great metric to have

I'm not sure it is either, how would you like to proceed with regard to this 
ticket?  If we don't want to add the metric then it should be possible to make 
ClientMetricsTest not check them for v2.

bq. client_request_metrics_test.TestClientRequestMetrics is failing simply 
because the metric isn't initialized until contention occurs

It turns out that v1 actually [does the same 
thing|https://github.com/apache/cassandra/blob/trunk/src/java/org/apache/cassandra/service/StorageProxy.java#L442],
 it's just that v2 doesn't encounter contention, so it's probably more correct 
to fix the test.  Unfortunately I don't think there's any way to detect the 
paxos variant from the dtest or ccm, so that may be difficult, and I don't 
think changing the test to cause contention will pan out either, so I'll to 
figure something out there.



> Paxos V2 does not update individual fields of readMetrics
> ---------------------------------------------------------
>
>                 Key: CASSANDRA-19046
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-19046
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Feature/Lightweight Transactions, Observability/Metrics
>            Reporter: Branimir Lambov
>            Assignee: Brandon Williams
>            Priority: Normal
>             Fix For: 4.1.x, 5.0-rc, 5.x
>
>
> As a result, {{ClientRequestMetricsTest.testPaxosStatement}} is failing with 
> {{{}paxos_variant: v2{}}}.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

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

Reply via email to