In Orchest multi-node configuration run interactive run containers in a single pod
ricklamers opened this issue · comments
Describe the problem this improvement solves
This will reduce latency for the multi-node configuration for interactive pipeline runs in the pipeline editor while developing the pipeline. Note pipeline runs that are part of job should continue to be scheduled as separate pods for pipeline steps to take advantage of cluster node availability.
Describe the solution you'd like
In Orchest multi-node configuration run interactive run containers in a single pod.
This is already the default in the single-node configuration.