Here is an update on running autopilot tests in the emulator on the cloud. In lp:~xnox/ubuntu-test-cases/touch-emulator, i've implemented a watchdog, to detect failed emulator startup, which will strip and retry upto 3 times. This results in significantly less "stuck" instances.
On the charm side, in lp:~xnox/charms/trusty/jemjem/trunk/ i've added scripts that help running tests across multiple workers. In practice, something like jenkins should be used to schedule execution on the nodes, but until then I have simple GNU parallel based scripts to execute full set of tests. Here is a quick tutorial on how to use them: $ juju bootstrap --constraints "cpu-cores=2 mem=2G arch=i386" $ juju deploy cs:~xnox/trusty/jemjem -n 9 ... wait for all emulators to be provisioned, takes approx. 21 minutes. In the mean time: $ bzr branch lp:~xnox/charms/trusty/jemjem/trunk jemjem $ ./script/get-ips # Once all instances are started: $ ./scripts/run-all.sh That last command probably best to be started in a GNU screen session or some such, as it takes about 1h 20m to complete. So, in a mere 1h40m from running "juju bootstrap" all autopilot tests can be executed (this is excluding unity8 tests, which I didn't get around to running yet). At the end, output/ will have raw console stdout/stderr from the run of each test, whilst clientlogs/ has the phablet-test-run results. I have now uploaded all of these for the image #119 at: http://people.canonical.com/~xnox/touch/ I guess I should have another run of image #120... =) -- Regards, Dimitri. -- Mailing list: https://launchpad.net/~ubuntu-phone Post to : ubuntu-phone@lists.launchpad.net Unsubscribe : https://launchpad.net/~ubuntu-phone More help : https://help.launchpad.net/ListHelp