Run stage regardless if previous jobs fail

Looking for a way to run a stage regardless if previous stages have failed or not.

We have parallel jobs in a test stage uploading coverage files to an s3 bucket, and a downstream stage with a job which syncs, sums, and uploads the coverage to code climate, then deletes the shared s3 bucket. The issue is that this downstream stage is canceled if any of the upstream test jobs fail (which is fine for not uploading the coverage report), but we can’t clean up the s3 bucket after the build

It sounds to me that allowing all of the jobs belonging to the first build stage to fail would achieve what you want.