id | title |
---|---|
Guide.ParallelTestExecution |
Parallel Test Execution |
Detox can leverage multi worker support of JS test runners (Jest, AVA, etc.).
By default detox test
will run the test runner with one worker (it will pass --maxWorkers=1
to Jest cli, Mocha is unaffected). Worker count can be controlled by adding --workers n
to detox test
, read more in detox-cli section.
Simulators/emulators run on a different process, outside of node, and require some sort of lock mechanism to make sure only one process controlls a simulator in a given time. Therefore, Detox 7.4.0 introduced device.registry.state.lock
, a lock file controlled by Detox, that registers all in-use simulators.
The lock file location is determined by the OS, and defined here.
MacOS:
~/Library/Detox/device.registry.state.lock
Linux:
~/.local/share/Detox/device.registry.state.lock
Windows:
$LOCALAPPDATA/data/Detox/device.registry.state.lock
or
$USERPROFILE/Application Data/Detox/device.registry.state.lock
While running with multiple workers, Detox might not have an available simulator for every worker.
If no simulator is available for that worker, the worker will create one with the name {name}-Detox
.
Each worker is responsible of removing the deviceId from the list in device.registry.state.lock
. Exiting a test runner abruptly (using ctrl+c / ⌘+c) will not give the worker a chance to deregister the device from the lock file, resulting in an inconsistent state, which ca result in creation of unnecessary new simulators.
-
detox-cli makes sure
device.registry.state.lock
is cleaned whenever it executes. -
If you use Detox without detox-cli make sure you delete or reset the lock file before running tests.
echo "[]" > ~/Library/Detox/device.registry.state.lock