hosting-cli(gcp): add --max-instances, --allow-unauthenticated, --env, --envfile

packages/reflex-hosting-cli/src/reflex_cli/v2/gcp.py

@@ -7,6 +7,7 @@
 tree is never modified. The script reads its parameters from environment
 variables (GCP_PROJECT, GCP_REGION, SERVICE_NAME, AR_REPO, VERSION,
 CLOUD_RUN_CPU, CLOUD_RUN_MEMORY, CLOUD_RUN_MIN_INSTANCES,
+CLOUD_RUN_MAX_INSTANCES, CLOUD_RUN_ALLOW_UNAUTHENTICATED,
 REFLEX_CLOUDBUILD_YAML).
 """
 
@@ -41,6 +42,8 @@
 ENV_CPU = "CLOUD_RUN_CPU"
 ENV_MEMORY = "CLOUD_RUN_MEMORY"
 ENV_MIN_INSTANCES = "CLOUD_RUN_MIN_INSTANCES"
+ENV_MAX_INSTANCES = "CLOUD_RUN_MAX_INSTANCES"
+ENV_ALLOW_UNAUTHENTICATED = "CLOUD_RUN_ALLOW_UNAUTHENTICATED"
 # Path to the Cloud Build config file written by the CLI. The rewritten
 # deploy script references it as ``--config="${REFLEX_CLOUDBUILD_YAML}"``.
 ENV_REFLEX_CLOUDBUILD_YAML = "REFLEX_CLOUDBUILD_YAML"
@@ -162,6 +165,21 @@
     type=click.IntRange(min=0),
     help="Minimum number of Cloud Run instances to keep warm (sets CLOUD_RUN_MIN_INSTANCES). Set to 0 to scale to zero.",
 )
+@click.option(
+    "--max-instances",
+    "max_instances",
+    default=100,
+    show_default=True,
+    type=click.IntRange(min=1),
+    help="Maximum number of Cloud Run instances during autoscaling (sets CLOUD_RUN_MAX_INSTANCES). Caps cost under traffic spikes.",
+)
+@click.option(
+    "--allow-unauthenticated/--no-allow-unauthenticated",
+    "allow_unauthenticated",
+    default=True,
+    show_default=True,
+    help="Whether to make the Cloud Run service publicly reachable (sets CLOUD_RUN_ALLOW_UNAUTHENTICATED). Use --no-allow-unauthenticated for internal / IAP-fronted services; callers will then need a roles/run.invoker IAM binding.",
+)
 @click.option(
     "--source",
     "source_dir",
@@ -199,6 +217,8 @@ def deploy_command(
     cpu: str,
     memory: str,
     min_instances: int,
+    max_instances: int,
+    allow_unauthenticated: bool,
     source_dir: str,
     token: str | None,
     interactive: bool,
@@ -226,6 +246,11 @@ def deploy_command(
     if not gcp_project:
         console.error("--gcp-project is required when using --gcp.")
         raise click.exceptions.Exit(2)
+    if max_instances < min_instances:
+        console.error(
+            f"--max-instances ({max_instances}) must be >= --min-instances ({min_instances})."
+        )
+        raise click.exceptions.Exit(2)
 
     authenticated_client = hosting.get_authenticated_client(
         token=token, interactive=interactive
@@ -284,6 +309,8 @@ def deploy_command(
         ENV_CPU: cpu,
         ENV_MEMORY: memory,
         ENV_MIN_INSTANCES: str(min_instances),
+        ENV_MAX_INSTANCES: str(max_instances),
+        ENV_ALLOW_UNAUTHENTICATED: "true" if allow_unauthenticated else "false",
     }
 
     console.info("Received deploy manifest from Reflex.")

tests/units/reflex_cli/v2/test_gcp.py

@@ -206,6 +206,116 @@ def test_gcp_deploy_resource_flags_have_defaults(mocker: MockFixture, tmp_path:
     assert env_overrides["CLOUD_RUN_MIN_INSTANCES"] == "1"
 
 
+def test_gcp_deploy_forwards_max_instances(mocker: MockFixture, tmp_path: Path):
+    """--max-instances threads through to CLOUD_RUN_MAX_INSTANCES."""
+    run_mock = _patch_environment(mocker)
+    _mock_manifest_response(mocker)
+
+    result = runner.invoke(
+        hosting_cli,
+        [
+            "deploy",
+            "--gcp",
+            "--gcp-project",
+            "p",
+            "--source",
+            str(tmp_path),
+            "--max-instances",
+            "42",
+        ],
+        input="y\n",
+    )
+
+    assert result.exit_code == 0, result.output
+    env_overrides = run_mock.call_args.kwargs["env_overrides"]
+    assert env_overrides["CLOUD_RUN_MAX_INSTANCES"] == "42"
+
+
+def test_gcp_deploy_max_instances_default(mocker: MockFixture, tmp_path: Path):
+    """Default --max-instances is 100, matching Cloud Run's own default."""
+    run_mock = _patch_environment(mocker)
+    _mock_manifest_response(mocker)
+
+    result = runner.invoke(
+        hosting_cli,
+        ["deploy", "--gcp", "--gcp-project", "p", "--source", str(tmp_path)],
+        input="y\n",
+    )
+
+    assert result.exit_code == 0, result.output
+    env_overrides = run_mock.call_args.kwargs["env_overrides"]
+    assert env_overrides["CLOUD_RUN_MAX_INSTANCES"] == "100"
+
+
+def test_gcp_deploy_rejects_max_less_than_min(mocker: MockFixture, tmp_path: Path):
+    """--max-instances < --min-instances is caught at the CLI, not inside gcloud."""
+    run_mock = _patch_environment(mocker)
+    _mock_manifest_response(mocker)
+
+    result = runner.invoke(
+        hosting_cli,
+        [
+            "deploy",
+            "--gcp",
+            "--gcp-project",
+            "p",
+            "--source",
+            str(tmp_path),
+            "--min-instances",
+            "5",
+            "--max-instances",
+            "3",
+        ],
+    )
+
+    assert result.exit_code == 2
+    assert "max-instances" in result.output.lower()
+    assert "min-instances" in result.output.lower()
+    assert run_mock.call_count == 0
+
+
+def test_gcp_deploy_allow_unauthenticated_defaults_true(
+    mocker: MockFixture, tmp_path: Path
+):
+    """Default is --allow-unauthenticated (public service), matching prior behavior."""
+    run_mock = _patch_environment(mocker)
+    _mock_manifest_response(mocker)
+
+    result = runner.invoke(
+        hosting_cli,
+        ["deploy", "--gcp", "--gcp-project", "p", "--source", str(tmp_path)],
+        input="y\n",
+    )
+
+    assert result.exit_code == 0, result.output
+    env_overrides = run_mock.call_args.kwargs["env_overrides"]
+    assert env_overrides["CLOUD_RUN_ALLOW_UNAUTHENTICATED"] == "true"
+
+
+def test_gcp_deploy_no_allow_unauthenticated(mocker: MockFixture, tmp_path: Path):
+    """--no-allow-unauthenticated produces the 'false' value."""
+    run_mock = _patch_environment(mocker)
+    _mock_manifest_response(mocker)
+
+    result = runner.invoke(
+        hosting_cli,
+        [
+            "deploy",
+            "--gcp",
+            "--gcp-project",
+            "p",
+            "--source",
+            str(tmp_path),
+            "--no-allow-unauthenticated",
+        ],
+        input="y\n",
+    )
+
+    assert result.exit_code == 0, result.output
+    env_overrides = run_mock.call_args.kwargs["env_overrides"]
+    assert env_overrides["CLOUD_RUN_ALLOW_UNAUTHENTICATED"] == "false"
+
+
 def test_gcp_deploy_rejects_negative_min_instances(mocker: MockFixture, tmp_path: Path):
     """--min-instances is IntRange(min=0); negative values fail at the CLI layer."""
     run_mock = _patch_environment(mocker)