Re: [HTCondor-users] condor_submit next_job_start_delay is not working

[Todd L Miller <tlmiller@xxxxxxxxxxx>]

> make sense? any better way of handling it?

	If you're OK with slowing all of your DAG's job submissions down, 
you could try using DAGMan's -maxidle option when submitting the DAG.  If 
all your EC2 jobs (or all the EC2 jobs with the problematic instance type) 
are in one node, a small subset of nodes, you could convert that subgraph 
of the DAG into a sub-DAG and set /its/ -maxidle.  Setting -maxidle 1, for 
example, would guarantee that you're only ever asking EC2 for one of the 
problematic resources at a time.

- Todd Miller