If the slot is full, Flink task submission will fail. How can I make the task enter the waiting state? Do I need support from other scheduling tools?

👋 Which kind of cluster deployment do you run? I suppose something like Standalone (new TMs cannot be provisioned on request) cand Session mode (you want to run some, for example as hoc jobs) If I understand your question correctly: I would try to resolve such problem with kind of Retrier on submitter side and be able to handle the Cluster is full case.

yes, I use the Standalone deployment, thank you