@Jthomas I don’t not have any benchmarks unfortunately. However, in general, we see faster execution time for jobs using a Docker executor vs. a machine executor when one of two conditions are true:
a convenience image is used. Convenience images are more likely to have layers already be cached on the host on which your job is running which reduces the need to pull all layers from scratch which can be slow.
A custom docker image is used. This lets you bundle your dependencies and any other relevant code into the image itself instead of having to spend time at the beginning of pulling those items down at the start of a job.
It is not possible to run Docker with ARM even with the version: default. I am getting:
“Job was rejected because this version of Docker is not supported on this resource class. Try removing the version of Docker in your .circleci/config.yml and re-running”
Hello! @sebastian-lerner
Had the same error message: Job was rejected because this version of Docker is not supported on this resource class. Try removing the version of Docker in your .circleci/config.yml and re-running