[
https://issues.apache.org/jira/browse/CASSANDRA-9526?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14711421#comment-14711421
]
Ariel Weisberg commented on CASSANDRA-9526:
-------------------------------------------
[Rebased to 2.2 and
squashed.|https://github.com/apache/cassandra/compare/trunk...aweisberg:CASSANDRA-9526-squashed?expand=1]
[utests|https://cassci.datastax.com/view/Dev/view/aweisberg/job/aweisberg-CASSANDRA-9526-testall/1/]
and
[dtests|http://cassci.datastax.com/view/Dev/view/aweisberg/job/aweisberg-CASSANDRA-9526-dtest/]
The core submission hasn't changed. I corrected some style issues like newline
after brace. The logging changes look like an improvement to me, but the
failure detector is not an area of expertise for me yet. I also updated
CHANGES.txt. It seems like you don't take away any existing information or make
things chattier except in the case of interpret() which if trace is enabled
will always log now.
In the last case it logs two statements back to back. Maybe I should change
that to be a single statement with both the phi value and mean?
> Provide a JMX hook to monitor phi values in the FailureDetector
> ---------------------------------------------------------------
>
> Key: CASSANDRA-9526
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9526
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Ron Kuris
> Fix For: 2.0.x
>
> Attachments: Monitor-Phi-JMX.patch, Phi-Log-Debug-When-Close.patch,
> Tiny-Race-Condition.patch
>
>
> phi_convict_threshold can be tuned, but there's currently no way to monitor
> the phi values to see if you're getting close.
> The attached patch adds the ability to get these values via JMX.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
