modal 1.1.1.dev41__py3-none-any.whl → 1.1.2__py3-none-any.whl

modal/cli/programs/launch_instance_ssh.py

@@ -0,0 +1,94 @@
+# Copyright Modal Labs 2023
+# type: ignore
+import json
+import os
+import sys
+from typing import Any
+
+import rich
+import rich.panel
+import rich.rule
+
+import modal
+import modal.experimental
+
+# Passed by `modal launch` locally via CLI.
+args: dict[str, Any] = json.loads(os.environ.get("MODAL_LAUNCH_ARGS", "{}"))
+
+app = modal.App()
+
+image: modal.Image
+if args.get("image"):
+    image = modal.Image.from_registry(args.get("image"))
+else:
+    # Must be set to the same image builder version as the notebook base image.
+    os.environ["MODAL_IMAGE_BUILDER_VERSION"] = "2024.10"
+    image = modal.experimental.notebook_base_image(python_version="3.12")
+
+volume = (
+    modal.Volume.from_name(
+        args.get("volume"),
+        create_if_missing=True,
+    )
+    if args.get("volume")
+    else None
+)
+volumes = {"/workspace": volume} if volume else {}
+
+
+startup_script = """
+set -eu
+mkdir -p /run/sshd
+
+# Check if sshd is installed, install if not
+test -x /usr/sbin/sshd || (apt-get update && apt-get install -y openssh-server)
+
+# Change default working directory to /workspace
+echo "cd /workspace" >> /root/.profile
+
+mkdir -p /root/.ssh
+echo "$SSH_PUBLIC_KEY" >> /root/.ssh/authorized_keys
+/usr/sbin/sshd -D -e
+"""
+
+
+@app.local_entrypoint()
+def main():
+    if not os.environ.get("SSH_PUBLIC_KEY"):
+        raise ValueError("SSH_PUBLIC_KEY environment variable is not set")
+
+    sb = modal.Sandbox.create(
+        *("sh", "-c", startup_script),
+        app=app,
+        image=image,
+        cpu=args.get("cpu"),
+        memory=args.get("memory"),
+        gpu=args.get("gpu"),
+        timeout=args.get("timeout"),
+        volumes=volumes,
+        unencrypted_ports=[22],  # Forward SSH port
+        secrets=[modal.Secret.from_dict({"SSH_PUBLIC_KEY": os.environ.get("SSH_PUBLIC_KEY")})],
+    )
+    hostname, port = sb.tunnels()[22].tcp_socket
+    connection_cmd = f"ssh -A -p {port} root@{hostname}"
+
+    rich.print(
+        rich.rule.Rule(style="yellow"),
+        rich.panel.Panel(
+            f"""Your instance is ready! You can SSH into it using the following command:
+
+  [dim gray]>[/dim gray] [bold cyan]{connection_cmd}[/bold cyan]
+
+[italic]Details:[/italic]
+  • Name:    [magenta]{app.description}[/magenta]
+  • CPU:     [yellow]{args.get("cpu")} cores[/yellow]
+  • Memory:  [yellow]{args.get("memory")} MiB[/yellow]
+  • Timeout: [yellow]{args.get("timeout")} seconds[/yellow]
+  • GPU:     [green]{(args.get("gpu") or "N/A").upper()}[/green]""",
+            title="SSH Connection",
+            expand=False,
+        ),
+        rich.rule.Rule(style="yellow"),
+    )
+
+    sys.exit(0)  # Exit immediately to prevent "Timed out waiting for final apps log."

modal/cli/programs/run_marimo.py

@@ -0,0 +1,95 @@
+# Copyright Modal Labs 2025
+# type: ignore
+import json
+import os
+import secrets
+import socket
+import subprocess
+import threading
+import time
+import webbrowser
+from typing import Any
+
+from modal import App, Image, Queue, Secret, Volume, forward
+
+# Args injected by `modal launch` CLI.
+args: dict[str, Any] = json.loads(os.environ.get("MODAL_LAUNCH_ARGS", "{}"))
+
+app = App()
+
+image = Image.from_registry(args.get("image"), add_python=args.get("add_python")).uv_pip_install("marimo")
+
+# Optional host-filesystem mount (read-only snapshot of your project, useful for editing)
+if args.get("mount"):
+    image = image.add_local_dir(args["mount"], remote_path="/root/marimo/mount")
+
+# Optional persistent Modal volume
+volume = Volume.from_name(args["volume"], create_if_missing=True) if args.get("volume") else None
+volumes = {"/root/marimo/volume": volume} if volume else {}
+
+
+def _wait_for_port(url: str, q: Queue) -> None:
+    start = time.monotonic()
+    while True:
+        try:
+            with socket.create_connection(("localhost", 8888), timeout=30):
+                break
+        except OSError as exc:
+            if time.monotonic() - start > 30:
+                raise TimeoutError("marimo server did not start within 30 s") from exc
+            time.sleep(0.05)
+    q.put(url)
+
+
+@app.function(
+    image=image,
+    cpu=args.get("cpu"),
+    memory=args.get("memory"),
+    gpu=args.get("gpu"),
+    timeout=args.get("timeout"),
+    secrets=[Secret.from_dict({"MODAL_LAUNCH_ARGS": json.dumps(args)})],
+    volumes=volumes,
+    max_containers=1 if volume else None,
+)
+def run_marimo(q: Queue):
+    os.makedirs("/root/marimo", exist_ok=True)
+
+    # marimo supports token-based auth; generate one so only you can connect
+    token = secrets.token_urlsafe(12)
+
+    with forward(8888) as tunnel:
+        url = f"{tunnel.url}/?access_token={token}"
+        threading.Thread(target=_wait_for_port, args=(url, q), daemon=True).start()
+
+        print("\nmarimo on Modal, opening in browser …")
+        print(f"   -> {url}\n")
+
+        # Launch the headless edit server
+        subprocess.run(
+            [
+                "marimo",
+                "edit",
+                "--headless",  # don't open browser in container
+                "--host",
+                "0.0.0.0",  # bind all interfaces
+                "--port",
+                "8888",
+                "--token-password",
+                token,  # enable session-based auth
+                "--skip-update-check",
+                "/root/marimo",  # workspace directory
+            ],
+            env={**os.environ, "SHELL": "/bin/bash"},
+        )
+
+    q.put("done")
+
+
+@app.local_entrypoint()
+def main():
+    with Queue.ephemeral() as q:
+        run_marimo.spawn(q)
+        url = q.get()  # first message = connect URL
+        time.sleep(1)  # give server a heartbeat
+        webbrowser.open(url)
+        assert q.get() == "done"

modal/cli/queues.py

@@ -1,4 +1,5 @@
 # Copyright Modal Labs 2024
+from datetime import datetime
 from typing import Optional
 
 import typer
@@ -44,17 +45,22 @@ async def create(name: str, *, env: Optional[str] = ENV_OPTION):
 
 @queue_cli.command(name="delete", rich_help_panel="Management")
 @synchronizer.create_blocking
-async def delete(name: str, *, yes: bool = YES_OPTION, env: Optional[str] = ENV_OPTION):
+async def delete(
+    name: str,
+    *,
+    allow_missing: bool = Option(False, "--allow-missing", help="Don't error if the Queue doesn't exist."),
+    yes: bool = YES_OPTION,
+    env: Optional[str] = ENV_OPTION,
+):
     """Delete a named Queue and all of its data."""
-    # Lookup first to validate the name, even though delete is a staticmethod
-    await _Queue.from_name(name, environment_name=env).hydrate()
+    env = ensure_env(env)
     if not yes:
         typer.confirm(
             f"Are you sure you want to irrevocably delete the modal.Queue '{name}'?",
             default=False,
             abort=True,
         )
-    await _Queue.delete(name, environment_name=env)
+    await _Queue.objects.delete(name, environment_name=env, allow_missing=allow_missing)
 
 
 @queue_cli.command(name="list", rich_help_panel="Management")
@@ -62,22 +68,46 @@ async def delete(name: str, *, yes: bool = YES_OPTION, env: Optional[str] = ENV_
 async def list_(*, json: bool = False, env: Optional[str] = ENV_OPTION):
     """List all named Queues."""
     env = ensure_env(env)
-
-    max_total_size = 100_000
     client = await _Client.from_env()
-    request = api_pb2.QueueListRequest(environment_name=env, total_size_limit=max_total_size + 1)
-    response = await retry_transient_errors(client.stub.QueueList, request)
-
-    rows = [
-        (
-            q.name,
-            timestamp_to_localized_str(q.created_at, json),
-            str(q.num_partitions),
-            str(q.total_size) if q.total_size <= max_total_size else f">{max_total_size}",
+    max_total_size = 100_000  # Limit on the *Queue size* that we report
+
+    items: list[api_pb2.QueueListResponse.QueueInfo] = []
+
+    # Note that we need to continue using the gRPC API directly here rather than using Queue.objects.list.
+    # There is some metadata that historically appears in the CLI output (num_partitions, total_size) that
+    # doesn't make sense to transmit as hydration metadata, because the values can change over time and
+    # the metadata retrieved at hydration time could get stale. Alternatively, we could rewrite this using
+    # only public API by sequentially retrieving the queues and then querying their dynamic metadata, but
+    # that would require multiple round trips and would add lag to the CLI.
+    async def retrieve_page(created_before: float) -> bool:
+        max_page_size = 100
+        pagination = api_pb2.ListPagination(max_objects=max_page_size, created_before=created_before)
+        req = api_pb2.QueueListRequest(environment_name=env, pagination=pagination, total_size_limit=max_total_size)
+        resp = await retry_transient_errors(client.stub.QueueList, req)
+        items.extend(resp.queues)
+        return len(resp.queues) < max_page_size
+
+    finished = await retrieve_page(datetime.now().timestamp())
+    while True:
+        if finished:
+            break
+        finished = await retrieve_page(items[-1].metadata.creation_info.created_at)
+
+    queues = [_Queue._new_hydrated(item.queue_id, client, item.metadata, is_another_app=True) for item in items]
+
+    rows = []
+    for obj, resp_data in zip(queues, items):
+        info = await obj.info()
+        rows.append(
+            (
+                obj.name,
+                timestamp_to_localized_str(info.created_at.timestamp(), json),
+                info.created_by,
+                str(resp_data.num_partitions),
+                str(resp_data.total_size) if resp_data.total_size <= max_total_size else f">{max_total_size}",
+            )
         )
-        for q in response.queues
-    ]
-    display_table(["Name", "Created at", "Partitions", "Total size"], rows, json)
+    display_table(["Name", "Created at", "Created by", "Partitions", "Total size"], rows, json)
 
 
 @queue_cli.command(name="clear", rich_help_panel="Management")
@@ -119,7 +149,7 @@ async def peek(
 
 @queue_cli.command(name="len", rich_help_panel="Inspection")
 @synchronizer.create_blocking
-async def len(
+async def len_(
     name: str,
     partition: Optional[str] = PARTITION_OPTION,
     total: bool = Option(False, "-t", "--total", help="Compute the sum of the queue lengths across all partitions"),

modal/cli/secret.py

@@ -3,6 +3,7 @@ import json
 import os
 import platform
 import subprocess
+from datetime import datetime
 from pathlib import Path
 from tempfile import NamedTemporaryFile
 from typing import Optional
@@ -10,7 +11,7 @@ from typing import Optional
 import click
 import typer
 from rich.syntax import Syntax
-from typer import Argument
+from typer import Argument, Option
 
 from modal._output import make_console
 from modal._utils.async_utils import synchronizer
@@ -30,20 +31,45 @@ secret_cli = typer.Typer(name="secret", help="Manage secrets.", no_args_is_help=
 async def list_(env: Optional[str] = ENV_OPTION, json: bool = False):
     env = ensure_env(env)
     client = await _Client.from_env()
-    response = await retry_transient_errors(client.stub.SecretList, api_pb2.SecretListRequest(environment_name=env))
-    column_names = ["Name", "Created at", "Last used at"]
-    rows = []
 
-    for item in response.items:
+    items: list[api_pb2.SecretListItem] = []
+
+    # Note that we need to continue using the gRPC API directly here rather than using Secret.objects.list.
+    # There is some metadata that historically appears in the CLI output (last_used_at) that
+    # doesn't make sense to transmit as hydration metadata, because the value can change over time and
+    # the metadata retrieved at hydration time could get stale. Alternatively, we could rewrite this using
+    # only public API by sequentially retrieving the secrets and then querying their dynamic metadata, but
+    # that would require multiple round trips and would add lag to the CLI.
+    async def retrieve_page(created_before: float) -> bool:
+        max_page_size = 100
+        pagination = api_pb2.ListPagination(max_objects=max_page_size, created_before=created_before)
+        req = api_pb2.SecretListRequest(environment_name=env, pagination=pagination)
+        resp = await retry_transient_errors(client.stub.SecretList, req)
+        items.extend(resp.items)
+        return len(resp.items) < max_page_size
+
+    finished = await retrieve_page(datetime.now().timestamp())
+    while True:
+        if finished:
+            break
+        finished = await retrieve_page(items[-1].metadata.creation_info.created_at)
+
+    secrets = [_Secret._new_hydrated(item.secret_id, client, item.metadata, is_another_app=True) for item in items]
+
+    rows = []
+    for obj, resp_data in zip(secrets, items):
+        info = await obj.info()
         rows.append(
             [
-                item.label,
-                timestamp_to_localized_str(item.created_at, json),
-                timestamp_to_localized_str(item.last_used_at, json) if item.last_used_at else "-",
+                obj.name,
+                timestamp_to_localized_str(info.created_at.timestamp(), json),
+                info.created_by,
+                timestamp_to_localized_str(resp_data.last_used_at, json) if resp_data.last_used_at else "-",
             ]
         )
 
     env_part = f" in environment '{env}'" if env else ""
+    column_names = ["Name", "Created at", "Created by", "Last used at"]
     display_table(column_names, rows, json, title=f"Secrets{env_part}")
 
 
@@ -114,7 +140,11 @@ modal secret create my-credentials username=john password="$PASSWORD"
             raise click.UsageError(f"Non-string value for secret '{k}'")
 
     # Create secret
-    await _Secret.create_deployed(secret_name, env_dict, overwrite=force)
+    if force:
+        # TODO migrate this path once we support Secret.update()?
+        await _Secret._create_deployed(secret_name, env_dict, overwrite=force)
+    else:
+        await _Secret.objects.create(secret_name, env_dict)
 
     # Print code sample
     console = make_console()
@@ -132,26 +162,23 @@ def some_function():
     console.print(Syntax(example_code, "python"))
 
 
-@secret_cli.command("delete", help="Delete a named secret.")
+@secret_cli.command("delete", help="Delete a named Secret.")
 @synchronizer.create_blocking
 async def delete(
-    secret_name: str = Argument(help="Name of the modal.Secret to be deleted. Case sensitive"),
+    name: str = Argument(help="Name of the modal.Secret to be deleted. Case sensitive"),
+    *,
+    allow_missing: bool = Option(False, "--allow-missing", help="Don't error if the Secret doesn't exist."),
     yes: bool = YES_OPTION,
     env: Optional[str] = ENV_OPTION,
 ):
-    """TODO"""
     env = ensure_env(env)
-    secret = await _Secret.from_name(secret_name, environment_name=env).hydrate()
     if not yes:
         typer.confirm(
-            f"Are you sure you want to irrevocably delete the modal.Secret '{secret_name}'?",
+            f"Are you sure you want to irrevocably delete the modal.Secret '{name}'?",
             default=False,
             abort=True,
         )
-    client = await _Client.from_env()
-
-    # TODO: replace with API on `modal.Secret` when we add it
-    await client.stub.SecretDelete(api_pb2.SecretDeleteRequest(secret_id=secret.object_id))
+    await _Secret.objects.delete(name, environment_name=env, allow_missing=allow_missing)
 
 
 def get_text_from_editor(key) -> str:

modal/cli/volume.py

@@ -13,11 +13,9 @@ from typer import Argument, Option, Typer
 import modal
 from modal._output import OutputManager, ProgressHandler, make_console
 from modal._utils.async_utils import synchronizer
-from modal._utils.grpc_utils import retry_transient_errors
 from modal._utils.time_utils import timestamp_to_localized_str
 from modal.cli._download import _volume_download
 from modal.cli.utils import ENV_OPTION, YES_OPTION, display_table
-from modal.client import _Client
 from modal.environments import ensure_env
 from modal.volume import _AbstractVolumeUploadContextManager, _Volume
 from modal_proto import api_pb2
@@ -56,7 +54,7 @@ def create(
     version: Optional[int] = Option(default=None, help="VolumeFS version. (Experimental)"),
 ):
     env_name = ensure_env(env)
-    modal.Volume.create_deployed(name, environment_name=env, version=version)
+    modal.Volume.objects.create(name, environment_name=env, version=version)
     usage_code = f"""
 @app.function(volumes={{"/my_vol": modal.Volume.from_name("{name}")}})
 def some_func():
@@ -110,14 +108,13 @@ async def get(
 @synchronizer.create_blocking
 async def list_(env: Optional[str] = ENV_OPTION, json: Optional[bool] = False):
     env = ensure_env(env)
-    client = await _Client.from_env()
-    response = await retry_transient_errors(client.stub.VolumeList, api_pb2.VolumeListRequest(environment_name=env))
-    env_part = f" in environment '{env}'" if env else ""
-    column_names = ["Name", "Created at"]
+    volumes = await _Volume.objects.list(environment_name=env)
     rows = []
-    for item in response.items:
-        rows.append([item.label, timestamp_to_localized_str(item.created_at, json)])
-    display_table(column_names, rows, json, title=f"Volumes{env_part}")
+    for obj in volumes:
+        info = await obj.info()
+        rows.append((info.name, timestamp_to_localized_str(info.created_at.timestamp(), json), info.created_by))
+
+    display_table(["Name", "Created at", "Created by"], rows, json)
 
 
 @volume_cli.command(
@@ -277,25 +274,26 @@ async def cp(
 
 @volume_cli.command(
     name="delete",
-    help="Delete a named, persistent modal.Volume.",
+    help="Delete a named Volume and all of its data.",
     rich_help_panel="Management",
 )
 @synchronizer.create_blocking
 async def delete(
-    volume_name: str = Argument(help="Name of the modal.Volume to be deleted. Case sensitive"),
+    name: str = Argument(help="Name of the modal.Volume to be deleted. Case sensitive"),
+    *,
+    allow_missing: bool = Option(False, "--allow-missing", help="Don't error if the Volume doesn't exist."),
     yes: bool = YES_OPTION,
     env: Optional[str] = ENV_OPTION,
 ):
-    # Lookup first to validate the name, even though delete is a staticmethod
-    await _Volume.from_name(volume_name, environment_name=env).hydrate()
+    env = ensure_env(env)
     if not yes:
         typer.confirm(
-            f"Are you sure you want to irrevocably delete the modal.Volume '{volume_name}'?",
+            f"Are you sure you want to irrevocably delete the modal.Volume '{name}'?",
             default=False,
             abort=True,
         )
 
-    await _Volume.delete(volume_name, environment_name=env)
+    await _Volume.objects.delete(name, environment_name=env, allow_missing=allow_missing)
 
 
 @volume_cli.command(

modal/client.pyi

@@ -29,11 +29,7 @@ class _Client:
     _snapshotted: bool
 
     def __init__(
-        self,
-        server_url: str,
-        client_type: int,
-        credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.1.1.dev41",
+        self, server_url: str, client_type: int, credentials: typing.Optional[tuple[str, str]], version: str = "1.1.2"
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.
@@ -160,11 +156,7 @@ class Client:
     _snapshotted: bool
 
     def __init__(
-        self,
-        server_url: str,
-        client_type: int,
-        credentials: typing.Optional[tuple[str, str]],
-        version: str = "1.1.1.dev41",
+        self, server_url: str, client_type: int, credentials: typing.Optional[tuple[str, str]], version: str = "1.1.2"
     ):
         """mdmd:hidden
         The Modal client object is not intended to be instantiated directly by users.

modal/cls.py

@@ -4,7 +4,7 @@ import inspect
 import os
 import typing
 from collections.abc import Collection
-from typing import Any, Callable, Optional, TypeVar, Union
+from typing import Any, Callable, Optional, Sequence, TypeVar, Union
 
 from google.protobuf.message import Message
 from grpclib import GRPCError, Status
@@ -37,6 +37,7 @@ from .config import config
 from .exception import ExecutionError, InvalidError, NotFoundError
 from .gpu import GPU_T
 from .retries import Retries
+from .scheduler_placement import SchedulerPlacement
 from .secret import _Secret
 from .volume import _Volume
 
@@ -92,6 +93,8 @@ class _ServiceOptions:
     target_concurrent_inputs: Optional[int] = None
     batch_max_size: Optional[int] = None
     batch_wait_ms: Optional[int] = None
+    scheduler_placement: Optional[api_pb2.SchedulerPlacement] = None
+    cloud: Optional[str] = None
 
     def merge_options(self, new_options: "_ServiceOptions") -> "_ServiceOptions":
         """Implement protobuf-like MergeFrom semantics for this dataclass.
@@ -657,7 +660,8 @@ More information on class parameterization can be found here: https://modal.com/
             await resolver.load(self._class_service_function)
             self._hydrate(response.class_id, resolver.client, response.handle_metadata)
 
-        rep = f"Cls.from_name({app_name!r}, {name!r})"
+        environment_rep = f", environment_name={environment_name!r}" if environment_name else ""
+        rep = f"Cls.from_name({app_name!r}, {name!r}{environment_rep})"
         cls = cls._from_loader(_load_remote, rep, is_another_app=True, hydrate_lazily=True)
 
         class_service_name = f"{name}.*"  # special name of the base service function for the class
@@ -684,6 +688,8 @@ More information on class parameterization can be found here: https://modal.com/
         buffer_containers: Optional[int] = None,  # Additional containers to scale up while Function is active.
         scaledown_window: Optional[int] = None,  # Max amount of time a container can remain idle before scaling down.
         timeout: Optional[int] = None,
+        region: Optional[Union[str, Sequence[str]]] = None,  # Region or regions to run the function on.
+        cloud: Optional[str] = None,  # Cloud provider to run the function on. Possible values are aws, gcp, oci, auto.
         # The following parameters are deprecated
         concurrency_limit: Optional[int] = None,  # Now called `max_containers`
         container_idle_timeout: Optional[int] = None,  # Now called `scaledown_window`
@@ -722,6 +728,8 @@ More information on class parameterization can be found here: https://modal.com/
         else:
             resources = None
 
+        scheduler_placement = SchedulerPlacement(region=region).proto if region else None
+
         if allow_concurrent_inputs is not None:
             deprecation_warning(
                 (2025, 5, 9),
@@ -757,6 +765,8 @@ More information on class parameterization can be found here: https://modal.com/
             buffer_containers=buffer_containers,
             scaledown_window=scaledown_window,
             timeout_secs=timeout,
+            scheduler_placement=scheduler_placement,
+            cloud=cloud,
             # Note: set both for backwards / forwards compatibility
             # But going forward `.with_concurrency` is the preferred method with distinct parameterization
             max_concurrent_inputs=allow_concurrent_inputs,

modal/cls.pyi

@@ -24,7 +24,7 @@ def _use_annotation_parameters(user_cls: type) -> bool: ...
 def _get_class_constructor_signature(user_cls: type) -> inspect.Signature: ...
 
 class _ServiceOptions:
-    """_ServiceOptions(secrets: Collection[modal.secret._Secret] = (), validated_volumes: Sequence[tuple[str, modal.volume._Volume]] = (), resources: Optional[modal_proto.api_pb2.Resources] = None, retry_policy: Optional[modal_proto.api_pb2.FunctionRetryPolicy] = None, max_containers: Optional[int] = None, buffer_containers: Optional[int] = None, scaledown_window: Optional[int] = None, timeout_secs: Optional[int] = None, max_concurrent_inputs: Optional[int] = None, target_concurrent_inputs: Optional[int] = None, batch_max_size: Optional[int] = None, batch_wait_ms: Optional[int] = None)"""
+    """_ServiceOptions(secrets: Collection[modal.secret._Secret] = (), validated_volumes: Sequence[tuple[str, modal.volume._Volume]] = (), resources: Optional[modal_proto.api_pb2.Resources] = None, retry_policy: Optional[modal_proto.api_pb2.FunctionRetryPolicy] = None, max_containers: Optional[int] = None, buffer_containers: Optional[int] = None, scaledown_window: Optional[int] = None, timeout_secs: Optional[int] = None, max_concurrent_inputs: Optional[int] = None, target_concurrent_inputs: Optional[int] = None, batch_max_size: Optional[int] = None, batch_wait_ms: Optional[int] = None, scheduler_placement: Optional[modal_proto.api_pb2.SchedulerPlacement] = None, cloud: Optional[str] = None)"""
 
     secrets: typing.Collection[modal.secret._Secret]
     validated_volumes: typing.Sequence[tuple[str, modal.volume._Volume]]
@@ -38,6 +38,8 @@ class _ServiceOptions:
     target_concurrent_inputs: typing.Optional[int]
     batch_max_size: typing.Optional[int]
     batch_wait_ms: typing.Optional[int]
+    scheduler_placement: typing.Optional[modal_proto.api_pb2.SchedulerPlacement]
+    cloud: typing.Optional[str]
 
     def merge_options(self, new_options: _ServiceOptions) -> _ServiceOptions:
         """Implement protobuf-like MergeFrom semantics for this dataclass.
@@ -60,6 +62,8 @@ class _ServiceOptions:
         target_concurrent_inputs: typing.Optional[int] = None,
         batch_max_size: typing.Optional[int] = None,
         batch_wait_ms: typing.Optional[int] = None,
+        scheduler_placement: typing.Optional[modal_proto.api_pb2.SchedulerPlacement] = None,
+        cloud: typing.Optional[str] = None,
     ) -> None:
         """Initialize self.  See help(type(self)) for accurate signature."""
         ...
@@ -395,6 +399,8 @@ class _Cls(modal._object._Object):
         buffer_containers: typing.Optional[int] = None,
         scaledown_window: typing.Optional[int] = None,
         timeout: typing.Optional[int] = None,
+        region: typing.Union[str, typing.Sequence[str], None] = None,
+        cloud: typing.Optional[str] = None,
         concurrency_limit: typing.Optional[int] = None,
         container_idle_timeout: typing.Optional[int] = None,
         allow_concurrent_inputs: typing.Optional[int] = None,
@@ -559,6 +565,8 @@ class Cls(modal.object.Object):
         buffer_containers: typing.Optional[int] = None,
         scaledown_window: typing.Optional[int] = None,
         timeout: typing.Optional[int] = None,
+        region: typing.Union[str, typing.Sequence[str], None] = None,
+        cloud: typing.Optional[str] = None,
         concurrency_limit: typing.Optional[int] = None,
         container_idle_timeout: typing.Optional[int] = None,
         allow_concurrent_inputs: typing.Optional[int] = None,

modal/config.py

@@ -94,7 +94,7 @@ from google.protobuf.empty_pb2 import Empty
 from modal_proto import api_pb2
 
 from ._utils.logger import configure_logger
-from .exception import InvalidError
+from .exception import InvalidError, NotFoundError
 
 DEFAULT_SERVER_URL = "https://api.modal.com"
 
@@ -158,15 +158,15 @@ def _config_active_profile() -> str:
         return "default"
 
 
-def config_set_active_profile(env: str) -> None:
+def config_set_active_profile(profile: str) -> None:
     """Set the user's active modal profile by writing it to the `.modal.toml` file."""
-    if env not in _user_config:
-        raise KeyError(env)
+    if profile not in _user_config:
+        raise NotFoundError(f"No profile named '{profile}' found in {user_config_path}")
 
-    for key, values in _user_config.items():
-        values.pop("active", None)
+    for profile_data in _user_config.values():
+        profile_data.pop("active", None)
 
-    _user_config[env]["active"] = True
+    _user_config[profile]["active"] = True  # type: ignore
     _write_user_config(_user_config)