canmove, Confirmed users
58
edits
Line 30: | Line 30: | ||
====False Regressions==== | ====False Regressions==== | ||
Sometimes we have false regressions either due to the test infrastructure or an error in datazilla. For each regression, if you cannot reproduce, please check the actual builds. For b2gperf, the link for builds are here - http://selenium.qa.mtv2.mozilla.com:8080/view/B2G%20Perf/. You need VPN access to get the builds. | Sometimes we have false regressions either due to the test infrastructure or an error in datazilla. For each regression, if you cannot reproduce, please check the actual builds. For b2gperf, the link for builds are here - http://selenium.qa.mtv2.mozilla.com:8080/view/B2G%20Perf/. You need VPN access to get the builds. If you need access, ping stephend. | ||
You have to manually dig through the logs to match the build id and gaia / gecko revision. These are usually found by looking at the console output, and searching for "gecko_revision" and "gaia_revision" and make sure they match up with datazilla. | You have to manually dig through the logs to match the build id and gaia / gecko revision. These are usually found by looking at the console output, and searching for "gecko_revision" and "gaia_revision" and make sure they match up with datazilla. | ||
One big signal that it is a datazilla issue is if the histogram of runs on datazilla show large sways and is not a consistent number. If this is the case, investigate datazilla before investigating the actual bisection. | One big signal that it is a datazilla issue is if the histogram of runs on datazilla show large sways and is not a consistent number. If this is the case, investigate datazilla before investigating the actual bisection. | ||
====Other Methods for Getting Revision ID Ranges==== | ====Other Methods for Getting Revision ID Ranges==== |