search

puppetlabs / gatling-puppet-load-test

34 stars 44 forks source link

lovejoy perf report, cut scale and ship it? #402

Open RandellP opened 4 years ago

johnduarte commented 4 years ago

The scale results are at least as good as Kearney no matter what the variance. IMHO we should document the failure to get reliable scale results in the report and note that all numbers were at least as good as Kearney to back up the claim that the scale expectations for Lovejoy should match those of Kearney.

billclaytor commented 4 years ago

Cold start may be a better, more conservative option, especially considering threadsafe jruby refresh. Cold start runs are in-progress. environment_timeout=0 runs are in-progress. Include some of the results in the report noting the variance.