Draft: merge Q2 submission support into main without recipes

src/srtctl/cli/do_sweep.py

@@ -556,13 +556,22 @@ def _run_post_eval(self, stop_event: threading.Event) -> int:
                 env_to_set["EVAL_CONC"] = str(max(conc_list))
                 logger.info("Eval concurrency (max of %s): %s", conc_list, env_to_set["EVAL_CONC"])
 
+        bash_preamble = None
+        if self.config.setup_script:
+            script_path = f"/configs/{self.config.setup_script}"
+            bash_preamble = (
+                f"echo 'Running setup script: {script_path}' && "
+                f"if [ -f '{script_path}' ]; then bash '{script_path}'; else echo 'WARNING: {script_path} not found'; fi"
+            )
+
         proc = start_srun_process(
             command=cmd,
             nodelist=[self.runtime.nodes.head],
             output=str(eval_log),
             container_image=str(self.runtime.container_image),
             container_mounts=self.runtime.container_mounts,
             env_to_set=env_to_set,
+            bash_preamble=bash_preamble,
             het_group=self.runtime.nodes.het_group_for(self.runtime.nodes.head),
         )
 

src/srtctl/core/config.py

@@ -152,6 +152,21 @@ def resolve_config_with_defaults(user_config: dict[str, Any], cluster_config: di
         model["container"] = resolved_container
         logger.debug(f"Resolved container alias '{container}' -> '{resolved_container}'")
 
+    # Resolve extra_mount host path aliases through model_paths. This lets
+    # recipes mount secondary model assets by alias rather than cluster path.
+    extra_mounts = config.get("extra_mount", [])
+    if model_paths and extra_mounts:
+        resolved_mounts = []
+        for mount_spec in extra_mounts:
+            host_path, container_path = mount_spec.split(":", 1)
+            if host_path in model_paths:
+                resolved_host = model_paths[host_path]
+                resolved_mounts.append(f"{resolved_host}:{container_path}")
+                logger.debug(f"Resolved extra_mount alias '{host_path}' -> '{resolved_host}'")
+            else:
+                resolved_mounts.append(mount_spec)
+        config["extra_mount"] = resolved_mounts
+
     # Apply reporting defaults (if not specified in user config)
     if "reporting" not in config and cluster_config.get("reporting"):
         config["reporting"] = cluster_config["reporting"]

tests/test_benchmarks.py

@@ -884,6 +884,36 @@ def test_successful_eval(self):
                     result = orch._run_post_eval(stop)
         assert result == 0
 
+    def test_post_eval_setup_script_preamble(self):
+        """setup_script runs before post-eval inside the eval container."""
+        import os
+        import threading
+        from dataclasses import replace
+        from unittest.mock import MagicMock, patch
+
+        orch = self._make_orchestrator()
+        orch.config = replace(orch.config, setup_script="install-trtllm-pip.sh")
+        stop = threading.Event()
+
+        mock_proc = MagicMock()
+        mock_proc.poll.return_value = 0
+        mock_proc.returncode = 0
+
+        captured_kwargs = {}
+
+        def capture_srun(**kwargs):
+            captured_kwargs.update(kwargs)
+            return mock_proc
+
+        with patch.dict(os.environ, {"EVAL_ONLY": "false"}, clear=False):
+            with patch("srtctl.cli.do_sweep.wait_for_port", return_value=True):
+                with patch("srtctl.cli.do_sweep.start_srun_process", side_effect=capture_srun):
+                    result = orch._run_post_eval(stop)
+
+        assert result == 0
+        assert "Running setup script: /configs/install-trtllm-pip.sh" in captured_kwargs["bash_preamble"]
+        assert "bash '/configs/install-trtllm-pip.sh'" in captured_kwargs["bash_preamble"]
+
     def test_eval_only_successful(self):
         """Returns 0 in eval-only mode when health check and eval succeed."""
         import os

tests/test_configs.py

@@ -642,6 +642,24 @@ def test_default_health_check_does_not_override_recipe(self):
 
         assert resolved["health_check"] == {"max_attempts": 720, "interval_seconds": 10}
 
+    def test_extra_mount_model_path_alias_resolution(self):
+        """extra_mount host paths can resolve through srtslurm.yaml model_paths."""
+        from srtctl.core.config import resolve_config_with_defaults
+
+        user_config = {
+            "name": "test",
+            "model": {"path": "/model", "container": "/c.sqsh", "precision": "fp8"},
+            "resources": {"gpu_type": "h100", "gpus_per_node": 8, "agg_nodes": 1},
+            "extra_mount": ["glm5-eagle:/eagle", "/literal/path:/literal"],
+        }
+
+        resolved = resolve_config_with_defaults(
+            user_config,
+            {"model_paths": {"glm5-eagle": "/models/glm5-eagle"}},
+        )
+
+        assert resolved["extra_mount"] == ["/models/glm5-eagle:/eagle", "/literal/path:/literal"]
+
     def test_cluster_sbatch_directives_are_not_treated_as_defaults(self):
         """srtslurm.yaml defaults must use default_sbatch_directives explicitly."""
         from srtctl.core.config import resolve_config_with_defaults