are there any docs on how data is passed from singer tap to target? im wondering if this is a legit idea… in a single k8s pod executed by airflow: • tap runs in container1 with a mounted config file • tap writes to a fifo file (shared volume between container 1 & 2) • target reads from fifo file • target runs in container2 with a mounted config file the result would be more generic, leaner, reusable images (only 1 image per tap/target, rather than 1 phat image for a whole meltano project)

Hi, @ben_marengo. I think you might have difficulty with the synchronous read and write from the same file, and at least as of today I don’t think there’s a smooth way to orchestrate this. In theory though, what you describe could probably work.