[
https://issues.apache.org/jira/browse/GEODE-6147?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16745370#comment-16745370
]
ASF subversion and git services commented on GEODE-6147:
--------------------------------------------------------
Commit d98d389fa775e09e5f68b9dc9a13acdea5b130fa in geode-benchmarks's branch
refs/heads/develop from Sean Goller
[ https://gitbox.apache.org/repos/asf?p=geode-benchmarks.git;h=d98d389 ]
GEODE-6147 - Fail benchmark task if average latency change is 5+% (#43)
* GEODE-6147 - Fail benchmark task if average latency change is 5+%
* Change ratio to difference
* Add difference calculation to ProbeResult
* Fail analyzeRun task if average latency difference is >= 5%
Authored-by: Sean Goller <[email protected]>
> Fail benchmark junit tests if their numbers deviate too far from a baseline
> ---------------------------------------------------------------------------
>
> Key: GEODE-6147
> URL: https://issues.apache.org/jira/browse/GEODE-6147
> Project: Geode
> Issue Type: Improvement
> Components: benchmarks
> Reporter: Dan Smith
> Assignee: Sean Goller
> Priority: Major
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> We want to fail tests if performance gets worse. We also want a signal to
> create a new baseline if performance gets better, so perhaps we should fail
> tests in that case as well.
> One way to do this would be to add a new system property to the
> TestRunners.defaultRunner() that takes in a baseline directory. The test
> itself could compare it's results with the baseline after running and fail if
> it deviates too far from the baseline.
> Using that, we could a new flag to the benchmark task to pass in that
> baseline and cause tests that deviate to fail.
> Eg
> ./gradlew benchmark -Pbaseline=/some/dir
> Acceptance:
> run_against_baseline.sh generates a failure report if benchmarks deviate too
> far from the baseline.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
