Spinning up an EMR cluster takes forever unfortunately, between the starting and the bootstrapping you’re in for a good 10 minutes at least.
I did encounter this ssl issue in the past and it seems to be transient with respect to the ec2 box you’re launching emr etl runner from, i.e. a restart usually solves the issue.
See Sluice fails on staging/archiving with OpenSSL error "Unsupported record version Unknown-0.0"