feat: add more resource specs to flux tracker job submit

examples/dist/state-machine-operator-dev.yaml

@@ -847,7 +847,7 @@ spec:
         control-plane: controller-manager
     spec:
       nodeSelector:
-        node.kubernetes.io/instance-type: m6a.4xlarge
+        node.kubernetes.io/instance-type: c7a.4xlarge
       containers:
       - args:
         - --metrics-bind-address=:8443

python/state_machine_operator/tracker/flux/tracker.py

@@ -65,14 +65,15 @@ def generate_flux_job(self, step, jobid):
         if not os.path.exists(entrypoint):
             utils.write_file(step.script, entrypoint)
 
-        # TODO expose more here?
-        # QUESTION: is a flux task == what we are using for procs?
+        exclusive = self.config.get("exclusive") in utils.true_values
+        num_tasks = max(1, self.config.get("tasks") or 1)
         jobspec = flux.job.JobspecV1.from_command(
             command=command,
-            # TODO add tasks here, right now assume == number of nodes
             num_nodes=step.nodes,
-            num_tasks=step.nodes,
+            num_tasks=num_tasks,
+            cores_per_task=step.cores_per_task,
             gpus_per_task=step.gpus,
+            exclusive=exclusive,
         )
 
         # Set user attribute we can later retrieve to identify group
@@ -147,6 +148,7 @@ def create_step(self, jobid):
             gpus=self.ngpus,
             workdir=workdir,
         )
+
         configfile = os.path.join(workdir, "app-config")
 
         if "script" in self.job_desc: