add debug output

Signed-off-by: Zoe Blevins <zblevins@nvidia.com>

Author: zsblevins

Makefile

@@ -718,6 +718,8 @@ local-up:
 # Uses a pre-built config profile (default: local-superpod.yaml).
 # Override INSTALL_CONFIG with a copied installer config when needed.
 HELM_TIMEOUT ?= 15m
+HELM_DEBUG ?=
+HELM_DEBUG_FLAG = $(if $(HELM_DEBUG),--helm-debug,)
 
 kind-up:
 	@echo "🚀 Deploying NVIDIA Config Manager with installer to Kind (config: $(INSTALL_CONFIG))..."
@@ -733,7 +735,7 @@ kind-up:
 		--install-envoy-gateway \
 		--install-cnpg-operator \
 		--install-cert-manager \
-		--helm-timeout $(HELM_TIMEOUT)
+		$(HELM_DEBUG_FLAG) --helm-timeout $(HELM_TIMEOUT)
 
 # Create Kind cluster, deploy NVIDIA Config Manager, and populate with mock topology.
 # The topology job is declared in the config profile's content.run_after_deploy,

installer/src/nv_config_manager_installer/cli.py

@@ -180,6 +180,7 @@ def generate_values(
 @click.option("--install-cert-manager", is_flag=True, help="Install cert-manager.")
 @click.option("--install-cnpg-operator", is_flag=True, help="Install CNPG operator.")
 @click.option("--helm-timeout", default="15m", help="Helm install/upgrade timeout.")
+@click.option("--helm-debug", is_flag=True, help="Enable Helm debug output during install/upgrade.")
 @click.option("--recreate-secrets", is_flag=True, help="Recreate existing K8s secrets.")
 @click.option("--dry-run", is_flag=True, help="Generate values only, skip helm install.")
 def deploy(
@@ -194,6 +195,7 @@ def deploy(
     install_cert_manager: bool,
     install_cnpg_operator: bool,
     helm_timeout: str,
+    helm_debug: bool,
     recreate_secrets: bool,
     dry_run: bool,
 ) -> None:
@@ -222,6 +224,7 @@ def deploy(
         install_cert_manager=install_cert_manager,
         install_cnpg_operator=install_cnpg_operator,
         helm_timeout=helm_timeout,
+        helm_debug=helm_debug,
         recreate_secrets=recreate_secrets,
         dry_run=dry_run,
     )

installer/src/nv_config_manager_installer/deployer.py

@@ -154,6 +154,7 @@ class DeployOptions:
     install_cert_manager: bool = False
     install_cnpg_operator: bool = False
     helm_timeout: str = "15m"
+    helm_debug: bool = False
     recreate_secrets: bool = False
     run_tests: bool = False
     dry_run: bool = False
@@ -2052,6 +2053,9 @@ def _helm_install(self) -> None:
             self.options.helm_timeout,
         ]
 
+        if self.options.helm_debug:
+            helm_args.append("--debug")
+
         if size_values.exists():
             helm_args.extend(["-f", str(size_values)])
 
@@ -2069,7 +2073,8 @@ def _helm_install(self) -> None:
                     f"WARNING: observability enabled but {observability_values} not found"
                 )
 
-        self.callback.on_log(f"Running: helm upgrade --install {release} ...")
+        debug_suffix = " --debug" if "--debug" in helm_args else ""
+        self.callback.on_log(f"Running: helm upgrade --install {release}{debug_suffix} ...")
         _run_logged(helm_args, step, self.callback, timeout=1200)
         self._finish_step(step)
 

installer/tests/test_cli.py

@@ -110,3 +110,4 @@ def test_deploy_help(self):
         assert "--load-kind" in result.output
         assert "--dry-run" in result.output
         assert "--helm-timeout" in result.output
+        assert "--helm-debug" in result.output

installer/tests/test_deployer.py

@@ -366,17 +366,20 @@ def test_defaults(self):
         assert opts.build_images is False
         assert opts.load_kind is False
         assert opts.helm_timeout == "15m"
+        assert opts.helm_debug is False
         assert opts.dry_run is False
 
     def test_custom_options(self):
         opts = DeployOptions(
             build_images=True,
             load_kind=True,
             kind_cluster="test-cluster",
+            helm_debug=True,
             dry_run=True,
         )
         assert opts.build_images is True
         assert opts.kind_cluster == "test-cluster"
+        assert opts.helm_debug is True
 
 
 class TestImageBuilds:
@@ -544,6 +547,82 @@ def fake_run_logged(cmd, step, callback, **kwargs):
         ] in logged_commands
 
 
+class TestHelmInstall:
+    def test_kind_deploy_does_not_enable_helm_debug_without_flag(self, monkeypatch, tmp_path):
+        logged_commands: list[list[str]] = []
+
+        def fake_run_logged(cmd, step, callback, **kwargs):
+            logged_commands.append(cmd)
+            return MagicMock(returncode=0, stdout="", stderr="")
+
+        monkeypatch.setattr("nv_config_manager_installer.deployer._run_logged", fake_run_logged)
+
+        config = _make_config()
+        config.cluster.airgapped = True
+        deployer = Deployer(
+            config,
+            DeployOptions(load_kind=True, chart_dir="deploy/helm"),
+            RecordingCallback(),
+        )
+        deployer._values_file = tmp_path / "values-generated.yaml"
+        deployer._values_file.write_text("global: {}\n")
+
+        deployer._helm_install()
+
+        helm_cmd = next(
+            cmd for cmd in logged_commands if cmd[:3] == ["helm", "upgrade", "--install"]
+        )
+        assert "--debug" not in helm_cmd
+
+    def test_helm_debug_can_be_enabled_without_kind(self, monkeypatch, tmp_path):
+        logged_commands: list[list[str]] = []
+
+        def fake_run_logged(cmd, step, callback, **kwargs):
+            logged_commands.append(cmd)
+            return MagicMock(returncode=0, stdout="", stderr="")
+
+        monkeypatch.setattr("nv_config_manager_installer.deployer._run_logged", fake_run_logged)
+
+        config = _make_config()
+        config.cluster.airgapped = True
+        deployer = Deployer(
+            config,
+            DeployOptions(helm_debug=True, chart_dir="deploy/helm"),
+            RecordingCallback(),
+        )
+        deployer._values_file = tmp_path / "values-generated.yaml"
+        deployer._values_file.write_text("global: {}\n")
+
+        deployer._helm_install()
+
+        helm_cmd = next(
+            cmd for cmd in logged_commands if cmd[:3] == ["helm", "upgrade", "--install"]
+        )
+        assert "--debug" in helm_cmd
+
+    def test_helm_debug_stays_off_by_default(self, monkeypatch, tmp_path):
+        logged_commands: list[list[str]] = []
+
+        def fake_run_logged(cmd, step, callback, **kwargs):
+            logged_commands.append(cmd)
+            return MagicMock(returncode=0, stdout="", stderr="")
+
+        monkeypatch.setattr("nv_config_manager_installer.deployer._run_logged", fake_run_logged)
+
+        config = _make_config()
+        config.cluster.airgapped = True
+        deployer = Deployer(config, DeployOptions(chart_dir="deploy/helm"), RecordingCallback())
+        deployer._values_file = tmp_path / "values-generated.yaml"
+        deployer._values_file.write_text("global: {}\n")
+
+        deployer._helm_install()
+
+        helm_cmd = next(
+            cmd for cmd in logged_commands if cmd[:3] == ["helm", "upgrade", "--install"]
+        )
+        assert "--debug" not in helm_cmd
+
+
 class TestContentHashing:
     def test_deterministic_hash(self):
         with tempfile.TemporaryDirectory() as d: