This is not for production yet; just get the raptor perf launch test (at minimum one app) running on the b2g-emulator on an AWS instance, using jenkins. Purpose is to get initial numbers to start and see if this platform will be consistent enough to establish a baseline. No need to use a docker image yet for this initial automation (except use Eli's docker-raptor image to host the visualization and help analyze the resulting numbers).

Note: The emulator boot-up is even slower than expected; going to have to increase the timeout in raptor for the emulator startup/reset (raptor/lib/suite/phase.js:30) to 240 seconds. Either that or add a RAPTOR_EMULATOR_TIMEOUT env variable so that it can be tweaked depending on platform. Probably a better solution, I will work on a patch for that soon.

Hmm, what if we duplicated the launch_test into something like emulator_launch_test? Then we can set the timeout in that file?

Also an update, I have jenkins running the launch_test on the emulator on AWS now, with the above timeout adjusted locally. Next step is get the visualization going; hoping that I can run the visualization in Eli's docker-raptor image locally and have it listen for test data on a forwarded AWS port. Either that or run the visualization docker container on AWS but tunnel the http port so I can view the data locally.

(In reply to :Eli Perelman from comment #2) > Hmm, what if we duplicated the launch_test into something like > emulator_launch_test? Then we can set the timeout in that file? We could, although we would need to do the same for each test that runs on the emulator though, as it is also the initial emulator boot-up that times out.

Thanks Eli, your suggestion makes sense, this way trying to keep as much emulator-specific code as possible out of the main raptor driver.

Thanks Eli. Merged: https://github.com/mozilla-b2g/gaia/commit/266c22986bf9ad1336aaacb915c4ebba151f7b58

I have the raptor perf launch test up and running on the b2g emulator on an AWS instance. Jenkins is used to download the emulator, checkout gaia at the same gaia rev, make raptor, run the test, and archive the results. Since this is not production, there's actually no need for the visualization here, as the purpose is to start and gather numbers (the raptor.log will be sufficient for now). Before reporting preliminary numbers, best to wait for Bug 1126567.

Didn't mean to mark it resolved.

Emulator_launch_test.js complete.