Using Fargate to autoscale is only creating 1 runner instance at a time

Some background - I need to create a set of prallel jobs that will all be created AND run at the same time. the parallel keyword can create jobs but due to the shared runner infrastructure there can be no guarantee the jobs will all be started at the same time.

So I followed the docs here: in order to have a set of runners that we can autoscale to as many as needed [50, 100, and on upwards] but the default configuration provided in the docs seems to be only creating one runner at a time, rather than the lots I need.

The extra kicker is that I’ve created these using the aws-cdk rather than through the console. The Gitlab docs don’t mention using a FargateService in order to create the desired number of instances, which is what I’m assuming is missing.

Has anyone hit this, or have any insights?