It works, I found the 2 tasks. The strange thing is that the flow are in a Failed state, but there is a task inside which stays in a running/blue state. I think those tasks/flows have failed because the pods were OOM by k8s. Here are the flow run ids, if it's of any use: • d2b9e49a-9c09-4c17-8037-266e29dea7fe • 8a7379de-013a-46d0-8c50-f97137362781 My guess is that it is related to my other thread on pod not being terminated after failures...

Maybe like this:

query {
  task_run (where: {state: {_eq: "Running"},
                    task: {tags: {_contains: "tag"}},
                    flow_run: {state: {_eq: "Failed"}}}) {
    id
    flow_run{
      name
      id
			state
    }
    task{
      name
      tags
    }
  }
}