I'm trying to understand the timing of the tasks. I'm using Sandbox only because I couldn't deploy it properly on AWS. For example, when using a

map_task

the UI shows simple execution of 1 task with access to the individual tasks. Which is great. I checked the logs of each task, after adding a print at the beginning of the execution, and it looks like they weren't running in parallel. The sandbox seem to have enough resources to run multiple pods.

have you set concurrency in the map task by any chance?

No, should I?

nope, not necessary. could you elaborate on why you think the tasks aren't running in parallel despite having enough resources?

honestly, it's a little hard to tell. I checked when the tasks started running. I'm using Flyte sandbox. By the K8s dashboard, it seems we still have some cores and memory to accommodate more than one task in parallel.

okay. can you try increasing the allotted memory to verify if more map tasks are being scheduled?