phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-21 08:03:09 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chore: remove build.py (#3869)
Some checks failed
SqlStore Integration Tests / test-postgres (3.12) (push) Failing after 0s
Details
Integration Auth Tests / test-matrix (oauth2_token) (push) Failing after 1s
Details
Integration Tests (Replay) / Integration Tests (, , , client=, ) (push) Failing after 3s
Details
SqlStore Integration Tests / test-postgres (3.13) (push) Failing after 6s
Details
Vector IO Integration Tests / test-matrix (push) Failing after 4s
Details
Python Package Build Test / build (3.13) (push) Failing after 1s
Details
Test Llama Stack Build / generate-matrix (push) Successful in 5s
Details
Test External Providers Installed via Module / test-external-providers-from-module (venv) (push) Has been skipped
Details
Test Llama Stack Build / build-single-provider (push) Failing after 3s
Details
Test Llama Stack Build / build-custom-container-distribution (push) Failing after 3s
Details
Test llama stack list-deps / generate-matrix (push) Successful in 4s
Details
Test llama stack list-deps / show-single-provider (push) Failing after 3s
Details
Test llama stack list-deps / list-deps-from-config (push) Failing after 3s
Details
API Conformance Tests / check-schema-compatibility (push) Successful in 11s
Details
Test External API and Providers / test-external (venv) (push) Failing after 4s
Details
Unit Tests / unit-tests (3.12) (push) Failing after 4s
Details
Test Llama Stack Build / build (push) Failing after 3s
Details
Unit Tests / unit-tests (3.13) (push) Failing after 4s
Details
Python Package Build Test / build (3.12) (push) Failing after 20s
Details
Test Llama Stack Build / build-ubi9-container-distribution (push) Failing after 23s
Details
Test llama stack list-deps / list-deps (push) Failing after 18s
Details
UI Tests / ui-tests (22) (push) Successful in 57s
Details
Pre-commit / pre-commit (push) Successful in 1m52s
Details

Browse source 
# What does this PR do?


## Test Plan
CI
This commit is contained in:
ehhuang 2025-10-20 16:28:15 -07:00 • committed by GitHub
parent 6a13a99e77
commit 444f6c88f3
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
15 changed files with 9 additions and 1306 deletions
Download patch file Download diff file Expand all files Collapse all files

519
llama_stack/cli/stack/_build.py
View file

@ -1,519 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import argparse
import importlib.resources
import json
import os
import shutil
import sys
import textwrap
from functools import lru_cache
from importlib.abc import Traversable
from pathlib import Path
import yaml
from prompt_toolkit import prompt
from prompt_toolkit.completion import WordCompleter
from prompt_toolkit.validation import Validator
from termcolor import colored, cprint
from llama_stack.cli.stack.utils import ImageType
from llama_stack.cli.table import print_table
from llama_stack.core.build import (
SERVER_DEPENDENCIES,
build_image,
get_provider_dependencies,
)
from llama_stack.core.configure import parse_and_maybe_upgrade_config
from llama_stack.core.datatypes import (
BuildConfig,
BuildProvider,
DistributionSpec,
Provider,
StackRunConfig,
)
from llama_stack.core.distribution import get_provider_registry
from llama_stack.core.external import load_external_apis
from llama_stack.core.resolver import InvalidProviderError
from llama_stack.core.stack import replace_env_vars
from llama_stack.core.storage.datatypes import (
InferenceStoreReference,
KVStoreReference,
ServerStoresConfig,
SqliteKVStoreConfig,
SqliteSqlStoreConfig,
SqlStoreReference,
StorageConfig,
)
from llama_stack.core.utils.config_dirs import DISTRIBS_BASE_DIR, EXTERNAL_PROVIDERS_DIR
from llama_stack.core.utils.dynamic import instantiate_class_type
from llama_stack.core.utils.exec import formulate_run_args, run_command
from llama_stack.core.utils.image_types import LlamaStackImageType
from llama_stack.providers.datatypes import Api
DISTRIBS_PATH = Path(__file__).parent.parent.parent / "distributions"
@lru_cache
def available_distros_specs() -> dict[str, BuildConfig]:
import yaml
distro_specs = {}
for p in DISTRIBS_PATH.rglob("*build.yaml"):
distro_name = p.parent.name
with open(p) as f:
build_config = BuildConfig(**yaml.safe_load(f))
distro_specs[distro_name] = build_config
return distro_specs
def run_stack_build_command(args: argparse.Namespace) -> None:
if args.list_distros:
return _run_distro_list_cmd()
if args.image_type == ImageType.VENV.value:
current_venv = os.environ.get("VIRTUAL_ENV")
image_name = args.image_name or current_venv
else:
image_name = args.image_name
if args.template:
cprint(
"The --template argument is deprecated. Please use --distro instead.",
color="red",
file=sys.stderr,
)
distro_name = args.template
else:
distro_name = args.distribution
if distro_name:
available_distros = available_distros_specs()
if distro_name not in available_distros:
cprint(
f"Could not find distribution {distro_name}. Please run `llama stack build --list-distros` to check out the available distributions",
color="red",
file=sys.stderr,
)
sys.exit(1)
build_config = available_distros[distro_name]
if args.image_type:
build_config.image_type = args.image_type
else:
cprint(
f"Please specify a image-type ({' | '.join(e.value for e in ImageType)}) for {distro_name}",
color="red",
file=sys.stderr,
)
sys.exit(1)
elif args.providers:
provider_list: dict[str, list[BuildProvider]] = dict()
for api_provider in args.providers.split(","):
if "=" not in api_provider:
cprint(
"Could not parse `--providers`. Please ensure the list is in the format api1=provider1,api2=provider2",
color="red",
file=sys.stderr,
)
sys.exit(1)
api, provider_type = api_provider.split("=")
providers_for_api = get_provider_registry().get(Api(api), None)
if providers_for_api is None:
cprint(
f"{api} is not a valid API.",
color="red",
file=sys.stderr,
)
sys.exit(1)
if provider_type in providers_for_api:
provider = BuildProvider(
provider_type=provider_type,
module=None,
)
provider_list.setdefault(api, []).append(provider)
else:
cprint(
f"{provider} is not a valid provider for the {api} API.",
color="red",
file=sys.stderr,
)
sys.exit(1)
distribution_spec = DistributionSpec(
providers=provider_list,
description=",".join(args.providers),
)
if not args.image_type:
cprint(
f"Please specify a image-type (container | venv) for {args.template}",
color="red",
file=sys.stderr,
)
sys.exit(1)
build_config = BuildConfig(image_type=args.image_type, distribution_spec=distribution_spec)
elif not args.config and not distro_name:
name = prompt(
"> Enter a name for your Llama Stack (e.g. my-local-stack): ",
validator=Validator.from_callable(
lambda x: len(x) > 0,
error_message="Name cannot be empty, please enter a name",
),
)
image_type = prompt(
"> Enter the image type you want your Llama Stack to be built as (use <TAB> to see options): ",
completer=WordCompleter([e.value for e in ImageType]),
complete_while_typing=True,
validator=Validator.from_callable(
lambda x: x in [e.value for e in ImageType],
error_message="Invalid image type. Use <TAB> to see options",
),
)
image_name = f"llamastack-{name}"
cprint(
textwrap.dedent(
"""
Llama Stack is composed of several APIs working together. Let's select
the provider types (implementations) you want to use for these APIs.
""",
),
color="green",
file=sys.stderr,
)
cprint("Tip: use <TAB> to see options for the providers.\n", color="green", file=sys.stderr)
providers: dict[str, list[BuildProvider]] = dict()
for api, providers_for_api in get_provider_registry().items():
available_providers = [x for x in providers_for_api.keys() if x not in ("remote", "remote::sample")]
if not available_providers:
continue
api_provider = prompt(
f"> Enter provider for API {api.value}: ",
completer=WordCompleter(available_providers),
complete_while_typing=True,
validator=Validator.from_callable(
lambda x: x in available_providers, # noqa: B023 - see https://github.com/astral-sh/ruff/issues/7847
error_message="Invalid provider, use <TAB> to see options",
),
)
string_providers = api_provider.split(" ")
for provider in string_providers:
providers.setdefault(api.value, []).append(BuildProvider(provider_type=provider))
description = prompt(
"\n > (Optional) Enter a short description for your Llama Stack: ",
default="",
)
distribution_spec = DistributionSpec(
providers=providers,
description=description,
)
build_config = BuildConfig(image_type=image_type, distribution_spec=distribution_spec)
else:
with open(args.config) as f:
try:
contents = yaml.safe_load(f)
contents = replace_env_vars(contents)
build_config = BuildConfig(**contents)
if args.image_type:
build_config.image_type = args.image_type
except Exception as e:
cprint(
f"Could not parse config file {args.config}: {e}",
color="red",
file=sys.stderr,
)
sys.exit(1)
if args.print_deps_only:
print(f"# Dependencies for {distro_name or args.config or image_name}")
normal_deps, special_deps, external_provider_dependencies = get_provider_dependencies(build_config)
normal_deps += SERVER_DEPENDENCIES
print(f"uv pip install {' '.join(normal_deps)}")
for special_dep in special_deps:
print(f"uv pip install {special_dep}")
for external_dep in external_provider_dependencies:
print(f"uv pip install {external_dep}")
return
try:
run_config = _run_stack_build_command_from_build_config(
build_config,
image_name=image_name,
config_path=args.config,
distro_name=distro_name,
)
except (Exception, RuntimeError) as exc:
import traceback
cprint(
f"Error building stack: {exc}",
color="red",
file=sys.stderr,
)
cprint("Stack trace:", color="red", file=sys.stderr)
traceback.print_exc()
sys.exit(1)
if run_config is None:
cprint(
"Run config path is empty",
color="red",
file=sys.stderr,
)
sys.exit(1)
if args.run:
config_dict = yaml.safe_load(run_config.read_text())
config = parse_and_maybe_upgrade_config(config_dict)
if config.external_providers_dir and not config.external_providers_dir.exists():
config.external_providers_dir.mkdir(exist_ok=True)
run_args = formulate_run_args(args.image_type, image_name or config.image_name)
run_args.extend([str(os.getenv("LLAMA_STACK_PORT", 8321)), "--config", str(run_config)])
run_command(run_args)
def _generate_run_config(
build_config: BuildConfig,
build_dir: Path,
image_name: str,
) -> Path:
"""
Generate a run.yaml template file for user to edit from a build.yaml file
"""
apis = list(build_config.distribution_spec.providers.keys())
distro_dir = DISTRIBS_BASE_DIR / image_name
storage = StorageConfig(
backends={
"kv_default": SqliteKVStoreConfig(
db_path=f"${{env.SQLITE_STORE_DIR:={distro_dir}}}/kvstore.db",
),
"sql_default": SqliteSqlStoreConfig(
db_path=f"${{env.SQLITE_STORE_DIR:={distro_dir}}}/sql_store.db",
),
},
stores=ServerStoresConfig(
metadata=KVStoreReference(
backend="kv_default",
namespace="registry",
),
inference=InferenceStoreReference(
backend="sql_default",
table_name="inference_store",
),
conversations=SqlStoreReference(
backend="sql_default",
table_name="openai_conversations",
),
),
)
run_config = StackRunConfig(
container_image=(image_name if build_config.image_type == LlamaStackImageType.CONTAINER.value else None),
image_name=image_name,
apis=apis,
providers={},
storage=storage,
external_providers_dir=build_config.external_providers_dir
if build_config.external_providers_dir
else EXTERNAL_PROVIDERS_DIR,
)
# build providers dict
provider_registry = get_provider_registry(build_config)
for api in apis:
run_config.providers[api] = []
providers = build_config.distribution_spec.providers[api]
for provider in providers:
pid = provider.provider_type.split("::")[-1]
p = provider_registry[Api(api)][provider.provider_type]
if p.deprecation_error:
raise InvalidProviderError(p.deprecation_error)
try:
config_type = instantiate_class_type(provider_registry[Api(api)][provider.provider_type].config_class)
except (ModuleNotFoundError, ValueError) as exc:
# HACK ALERT:
# This code executes after building is done, the import cannot work since the
# package is either available in the venv or container - not available on the host.
# TODO: use a "is_external" flag in ProviderSpec to check if the provider is
# external
cprint(
f"Failed to import provider {provider.provider_type} for API {api} - assuming it's external, skipping: {exc}",
color="yellow",
file=sys.stderr,
)
# Set config_type to None to avoid UnboundLocalError
config_type = None
if config_type is not None and hasattr(config_type, "sample_run_config"):
config = config_type.sample_run_config(__distro_dir__=f"~/.llama/distributions/{image_name}")
else:
config = {}
p_spec = Provider(
provider_id=pid,
provider_type=provider.provider_type,
config=config,
module=provider.module,
)
run_config.providers[api].append(p_spec)
run_config_file = build_dir / f"{image_name}-run.yaml"
with open(run_config_file, "w") as f:
to_write = json.loads(run_config.model_dump_json())
f.write(yaml.dump(to_write, sort_keys=False))
# Only print this message for non-container builds since it will be displayed before the
# container is built
# For non-container builds, the run.yaml is generated at the very end of the build process so it
# makes sense to display this message
if build_config.image_type != LlamaStackImageType.CONTAINER.value:
cprint(f"You can now run your stack with `llama stack run {run_config_file}`", color="green", file=sys.stderr)
return run_config_file
def _run_stack_build_command_from_build_config(
build_config: BuildConfig,
image_name: str | None = None,
distro_name: str | None = None,
config_path: str | None = None,
) -> Path | Traversable:
image_name = image_name or build_config.image_name
if build_config.image_type == LlamaStackImageType.CONTAINER.value:
if distro_name:
image_name = f"distribution-{distro_name}"
else:
if not image_name:
raise ValueError("Please specify an image name when building a container image without a template")
else:
if not image_name and os.environ.get("UV_SYSTEM_PYTHON"):
image_name = "__system__"
if not image_name:
raise ValueError("Please specify an image name when building a venv image")
# At this point, image_name should be guaranteed to be a string
if image_name is None:
raise ValueError("image_name should not be None after validation")
if distro_name:
build_dir = DISTRIBS_BASE_DIR / distro_name
build_file_path = build_dir / f"{distro_name}-build.yaml"
else:
if image_name is None:
raise ValueError("image_name cannot be None")
build_dir = DISTRIBS_BASE_DIR / image_name
build_file_path = build_dir / f"{image_name}-build.yaml"
os.makedirs(build_dir, exist_ok=True)
run_config_file = None
# Generate the run.yaml so it can be included in the container image with the proper entrypoint
# Only do this if we're building a container image and we're not using a template
if build_config.image_type == LlamaStackImageType.CONTAINER.value and not distro_name and config_path:
cprint("Generating run.yaml file", color="yellow", file=sys.stderr)
run_config_file = _generate_run_config(build_config, build_dir, image_name)
with open(build_file_path, "w") as f:
to_write = json.loads(build_config.model_dump_json(exclude_none=True))
f.write(yaml.dump(to_write, sort_keys=False))
# We first install the external APIs so that the build process can use them and discover the
# providers dependencies
if build_config.external_apis_dir:
cprint("Installing external APIs", color="yellow", file=sys.stderr)
external_apis = load_external_apis(build_config)
if external_apis:
# install the external APIs
packages = []
for _, api_spec in external_apis.items():
if api_spec.pip_packages:
packages.extend(api_spec.pip_packages)
cprint(
f"Installing {api_spec.name} with pip packages {api_spec.pip_packages}",
color="yellow",
file=sys.stderr,
)
return_code = run_command(["uv", "pip", "install", *packages])
if return_code != 0:
packages_str = ", ".join(packages)
raise RuntimeError(
f"Failed to install external APIs packages: {packages_str} (return code: {return_code})"
)
return_code = build_image(
build_config,
image_name,
distro_or_config=distro_name or config_path or str(build_file_path),
run_config=run_config_file.as_posix() if run_config_file else None,
)
if return_code != 0:
raise RuntimeError(f"Failed to build image {image_name}")
if distro_name:
# copy run.yaml from distribution to build_dir instead of generating it again
distro_path = importlib.resources.files("llama_stack") / f"distributions/{distro_name}/run.yaml"
run_config_file = build_dir / f"{distro_name}-run.yaml"
with importlib.resources.as_file(distro_path) as path:
shutil.copy(path, run_config_file)
cprint("Build Successful!", color="green", file=sys.stderr)
cprint(f"You can find the newly-built distribution here: {run_config_file}", color="blue", file=sys.stderr)
if build_config.image_type == LlamaStackImageType.VENV:
cprint(
"You can run the new Llama Stack distro (after activating "
+ colored(image_name, "cyan")
+ ") via: "
+ colored(f"llama stack run {run_config_file}", "blue"),
color="green",
file=sys.stderr,
)
elif build_config.image_type == LlamaStackImageType.CONTAINER:
cprint(
"You can run the container with: "
+ colored(
f"docker run -p 8321:8321 -v ~/.llama:/root/.llama localhost/{image_name} --port 8321", "blue"
),
color="green",
file=sys.stderr,
)
return distro_path
else:
return _generate_run_config(build_config, build_dir, image_name)
def _run_distro_list_cmd() -> None:
headers = [
"Distribution Name",
# "Providers",
"Description",
]
rows = []
for distro_name, spec in available_distros_specs().items():
rows.append(
[
distro_name,
# json.dumps(spec.distribution_spec.providers, indent=2),
spec.distribution_spec.description,
]
)
print_table(
rows,
headers,
separate_rows=True,
)

106
llama_stack/cli/stack/build.py
View file

@ -1,106 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import argparse
import textwrap
from llama_stack.cli.stack.utils import ImageType
from llama_stack.cli.subcommand import Subcommand
from llama_stack.log import get_logger
logger = get_logger(__name__, category="cli")
class StackBuild(Subcommand):
def __init__(self, subparsers: argparse._SubParsersAction):
super().__init__()
self.parser = subparsers.add_parser(
"build",
prog="llama stack build",
description="[DEPRECATED] Build a Llama stack container. This command is deprecated and will be removed in a future release. Use `llama stack list-deps <distro>' instead.",
formatter_class=argparse.ArgumentDefaultsHelpFormatter,
)
self._add_arguments()
self.parser.set_defaults(func=self._run_stack_build_command)
def _add_arguments(self):
self.parser.add_argument(
"--config",
type=str,
default=None,
help="Path to a config file to use for the build. You can find example configs in llama_stack.cores/**/build.yaml. If this argument is not provided, you will be prompted to enter information interactively",
)
self.parser.add_argument(
"--template",
type=str,
default=None,
help="""(deprecated) Name of the example template config to use for build. You may use `llama stack build --list-distros` to check out the available distributions""",
)
self.parser.add_argument(
"--distro",
"--distribution",
dest="distribution",
type=str,
default=None,
help="""Name of the distribution to use for build. You may use `llama stack build --list-distros` to check out the available distributions""",
)
self.parser.add_argument(
"--list-distros",
"--list-distributions",
action="store_true",
dest="list_distros",
default=False,
help="Show the available distributions for building a Llama Stack distribution",
)
self.parser.add_argument(
"--image-type",
type=str,
help="Image Type to use for the build. If not specified, will use the image type from the template config.",
choices=[e.value for e in ImageType],
default=None, # no default so we can detect if a user specified --image-type and override image_type in the config
)
self.parser.add_argument(
"--image-name",
type=str,
help=textwrap.dedent(
f"""[for image-type={"|".join(e.value for e in ImageType)}] Name of the virtual environment to use for
the build. If not specified, currently active environment will be used if found.
"""
),
default=None,
)
self.parser.add_argument(
"--print-deps-only",
default=False,
action="store_true",
help="Print the dependencies for the stack only, without building the stack",
)
self.parser.add_argument(
"--run",
action="store_true",
default=False,
help="Run the stack after building using the same image type, name, and other applicable arguments",
)
self.parser.add_argument(
"--providers",
type=str,
default=None,
help="Build a config for a list of providers and only those providers. This list is formatted like: api1=provider1,api2=provider2. Where there can be multiple providers per API.",
)
def _run_stack_build_command(self, args: argparse.Namespace) -> None:
logger.warning(
"The 'llama stack build' command is deprecated and will be removed in a future release. Please use 'llama stack list-deps'"
)
# always keep implementation completely silo-ed away from CLI so CLI
# can be fast to load and reduces dependencies
from ._build import run_stack_build_command
return run_stack_build_command(args)

2
llama_stack/cli/stack/stack.py
View file

@ -11,7 +11,6 @@ from llama_stack.cli.stack.list_stacks import StackListBuilds
from llama_stack.cli.stack.utils import print_subcommand_description
from llama_stack.cli.subcommand import Subcommand
from .build import StackBuild
from .list_apis import StackListApis
from .list_deps import StackListDeps
from .list_providers import StackListProviders
@ -41,7 +40,6 @@ class StackParser(Subcommand):
# Add sub-commands
StackListDeps.create(subparsers)
StackBuild.create(subparsers)
StackListApis.create(subparsers)
StackListProviders.create(subparsers)
StackRun.create(subparsers)

410
llama_stack/core/build_container.sh
View file

@ -1,410 +0,0 @@
#!/usr/bin/env bash
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
LLAMA_STACK_DIR=${LLAMA_STACK_DIR:-}
LLAMA_STACK_CLIENT_DIR=${LLAMA_STACK_CLIENT_DIR:-}
TEST_PYPI_VERSION=${TEST_PYPI_VERSION:-}
PYPI_VERSION=${PYPI_VERSION:-}
BUILD_PLATFORM=${BUILD_PLATFORM:-}
# This timeout (in seconds) is necessary when installing PyTorch via uv since it's likely to time out
# Reference: https://github.com/astral-sh/uv/pull/1694
UV_HTTP_TIMEOUT=${UV_HTTP_TIMEOUT:-500}
# mounting is not supported by docker buildx, so we use COPY instead
USE_COPY_NOT_MOUNT=${USE_COPY_NOT_MOUNT:-}
# Path to the run.yaml file in the container
RUN_CONFIG_PATH=/app/run.yaml
BUILD_CONTEXT_DIR=$(pwd)
set -euo pipefail
# Define color codes
RED='\033[0;31m'
NC='\033[0m' # No Color
# Usage function
usage() {
echo "Usage: $0 --image-name <image_name> --container-base <container_base> --normal-deps <pip_dependencies> [--run-config <run_config>] [--external-provider-deps <external_provider_deps>] [--optional-deps <special_pip_deps>]"
echo "Example: $0 --image-name llama-stack-img --container-base python:3.12-slim --normal-deps 'numpy pandas' --run-config ./run.yaml --external-provider-deps 'foo' --optional-deps 'bar'"
exit 1
}
# Parse arguments
image_name=""
container_base=""
normal_deps=""
external_provider_deps=""
optional_deps=""
run_config=""
distro_or_config=""
while [[ $# -gt 0 ]]; do
key="$1"
case "$key" in
--image-name)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --image-name requires a string value" >&2
usage
fi
image_name="$2"
shift 2
;;
--container-base)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --container-base requires a string value" >&2
usage
fi
container_base="$2"
shift 2
;;
--normal-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --normal-deps requires a string value" >&2
usage
fi
normal_deps="$2"
shift 2
;;
--external-provider-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --external-provider-deps requires a string value" >&2
usage
fi
external_provider_deps="$2"
shift 2
;;
--optional-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --optional-deps requires a string value" >&2
usage
fi
optional_deps="$2"
shift 2
;;
--run-config)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --run-config requires a string value" >&2
usage
fi
run_config="$2"
shift 2
;;
--distro-or-config)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --distro-or-config requires a string value" >&2
usage
fi
distro_or_config="$2"
shift 2
;;
*)
echo "Unknown option: $1" >&2
usage
;;
esac
done
# Check required arguments
if [[ -z "$image_name" || -z "$container_base" || -z "$normal_deps" ]]; then
echo "Error: --image-name, --container-base, and --normal-deps are required." >&2
usage
fi
CONTAINER_BINARY=${CONTAINER_BINARY:-docker}
CONTAINER_OPTS=${CONTAINER_OPTS:---progress=plain}
TEMP_DIR=$(mktemp -d)
SCRIPT_DIR=$(dirname "$(readlink -f "$0")")
source "$SCRIPT_DIR/common.sh"
add_to_container() {
output_file="$TEMP_DIR/Containerfile"
if [ -t 0 ]; then
printf '%s\n' "$1" >>"$output_file"
else
cat >>"$output_file"
fi
}
if ! is_command_available "$CONTAINER_BINARY"; then
printf "${RED}Error: ${CONTAINER_BINARY} command not found. Is ${CONTAINER_BINARY} installed and in your PATH?${NC}" >&2
exit 1
fi
if [[ $container_base == *"registry.access.redhat.com/ubi9"* ]]; then
add_to_container << EOF
FROM $container_base
WORKDIR /app
# We install the Python 3.12 dev headers and build tools so that any
# C-extension wheels (e.g. polyleven, faiss-cpu) can compile successfully.
RUN dnf -y update && dnf install -y iputils git net-tools wget \
vim-minimal python3.12 python3.12-pip python3.12-wheel \
python3.12-setuptools python3.12-devel gcc gcc-c++ make && \
ln -s /bin/pip3.12 /bin/pip && ln -s /bin/python3.12 /bin/python && dnf clean all
ENV UV_SYSTEM_PYTHON=1
RUN pip install uv
EOF
else
add_to_container << EOF
FROM $container_base
WORKDIR /app
RUN apt-get update && apt-get install -y \
iputils-ping net-tools iproute2 dnsutils telnet \
curl wget telnet git\
procps psmisc lsof \
traceroute \
bubblewrap \
gcc g++ \
&& rm -rf /var/lib/apt/lists/*
ENV UV_SYSTEM_PYTHON=1
RUN pip install uv
EOF
fi
# Add pip dependencies first since llama-stack is what will change most often
# so we can reuse layers.
if [ -n "$normal_deps" ]; then
read -ra pip_args <<< "$normal_deps"
quoted_deps=$(printf " %q" "${pip_args[@]}")
add_to_container << EOF
RUN uv pip install --no-cache $quoted_deps
EOF
fi
if [ -n "$optional_deps" ]; then
IFS='#' read -ra parts <<<"$optional_deps"
for part in "${parts[@]}"; do
read -ra pip_args <<< "$part"
quoted_deps=$(printf " %q" "${pip_args[@]}")
add_to_container <<EOF
RUN uv pip install --no-cache $quoted_deps
EOF
done
fi
if [ -n "$external_provider_deps" ]; then
IFS='#' read -ra parts <<<"$external_provider_deps"
for part in "${parts[@]}"; do
read -ra pip_args <<< "$part"
quoted_deps=$(printf " %q" "${pip_args[@]}")
add_to_container <<EOF
RUN uv pip install --no-cache $quoted_deps
EOF
add_to_container <<EOF
RUN python3 - <<PYTHON | uv pip install --no-cache -r -
import importlib
import sys
try:
package_name = '$part'.split('==')[0].split('>=')[0].split('<=')[0].split('!=')[0].split('<')[0].split('>')[0]
module = importlib.import_module(f'{package_name}.provider')
spec = module.get_provider_spec()
if hasattr(spec, 'pip_packages') and spec.pip_packages:
if isinstance(spec.pip_packages, (list, tuple)):
print('\n'.join(spec.pip_packages))
except Exception as e:
print(f'Error getting provider spec for {package_name}: {e}', file=sys.stderr)
PYTHON
EOF
done
fi
get_python_cmd() {
if is_command_available python; then
echo "python"
elif is_command_available python3; then
echo "python3"
else
echo "Error: Neither python nor python3 is installed. Please install Python to continue." >&2
exit 1
fi
}
if [ -n "$run_config" ]; then
# Copy the run config to the build context since it's an absolute path
cp "$run_config" "$BUILD_CONTEXT_DIR/run.yaml"
# Parse the run.yaml configuration to identify external provider directories
# If external providers are specified, copy their directory to the container
# and update the configuration to reference the new container path
python_cmd=$(get_python_cmd)
external_providers_dir=$($python_cmd -c "import yaml; config = yaml.safe_load(open('$run_config')); print(config.get('external_providers_dir') or '')")
external_providers_dir=$(eval echo "$external_providers_dir")
if [ -n "$external_providers_dir" ]; then
if [ -d "$external_providers_dir" ]; then
echo "Copying external providers directory: $external_providers_dir"
cp -r "$external_providers_dir" "$BUILD_CONTEXT_DIR/providers.d"
add_to_container << EOF
COPY providers.d /.llama/providers.d
EOF
fi
# Edit the run.yaml file to change the external_providers_dir to /.llama/providers.d
if [ "$(uname)" = "Darwin" ]; then
sed -i.bak -e 's|external_providers_dir:.*|external_providers_dir: /.llama/providers.d|' "$BUILD_CONTEXT_DIR/run.yaml"
rm -f "$BUILD_CONTEXT_DIR/run.yaml.bak"
else
sed -i 's|external_providers_dir:.*|external_providers_dir: /.llama/providers.d|' "$BUILD_CONTEXT_DIR/run.yaml"
fi
fi
# Copy run config into docker image
add_to_container << EOF
COPY run.yaml $RUN_CONFIG_PATH
EOF
fi
stack_mount="/app/llama-stack-source"
client_mount="/app/llama-stack-client-source"
install_local_package() {
local dir="$1"
local mount_point="$2"
local name="$3"
if [ ! -d "$dir" ]; then
echo "${RED}Warning: $name is set but directory does not exist: $dir${NC}" >&2
exit 1
fi
if [ "$USE_COPY_NOT_MOUNT" = "true" ]; then
add_to_container << EOF
COPY $dir $mount_point
EOF
fi
add_to_container << EOF
RUN uv pip install --no-cache -e $mount_point
EOF
}
if [ -n "$LLAMA_STACK_CLIENT_DIR" ]; then
install_local_package "$LLAMA_STACK_CLIENT_DIR" "$client_mount" "LLAMA_STACK_CLIENT_DIR"
fi
if [ -n "$LLAMA_STACK_DIR" ]; then
install_local_package "$LLAMA_STACK_DIR" "$stack_mount" "LLAMA_STACK_DIR"
else
if [ -n "$TEST_PYPI_VERSION" ]; then
# these packages are damaged in test-pypi, so install them first
add_to_container << EOF
RUN uv pip install --no-cache fastapi libcst
EOF
add_to_container << EOF
RUN uv pip install --no-cache --extra-index-url https://test.pypi.org/simple/ \
--index-strategy unsafe-best-match \
llama-stack==$TEST_PYPI_VERSION
EOF
else
if [ -n "$PYPI_VERSION" ]; then
SPEC_VERSION="llama-stack==${PYPI_VERSION}"
else
SPEC_VERSION="llama-stack"
fi
add_to_container << EOF
RUN uv pip install --no-cache $SPEC_VERSION
EOF
fi
fi
# remove uv after installation
add_to_container << EOF
RUN pip uninstall -y uv
EOF
# If a run config is provided, we use the llama stack CLI
if [[ -n "$run_config" ]]; then
add_to_container << EOF
ENTRYPOINT ["llama", "stack", "run", "$RUN_CONFIG_PATH"]
EOF
elif [[ "$distro_or_config" != *.yaml ]]; then
add_to_container << EOF
ENTRYPOINT ["llama", "stack", "run", "$distro_or_config"]
EOF
fi
# Add other require item commands genearic to all containers
add_to_container << EOF
RUN mkdir -p /.llama /.cache && chmod -R g+rw /.llama /.cache && (chmod -R g+rw /app 2>/dev/null || true)
EOF
printf "Containerfile created successfully in %s/Containerfile\n\n" "$TEMP_DIR"
cat "$TEMP_DIR"/Containerfile
printf "\n"
# Start building the CLI arguments
CLI_ARGS=()
# Read CONTAINER_OPTS and put it in an array
read -ra CLI_ARGS <<< "$CONTAINER_OPTS"
if [ "$USE_COPY_NOT_MOUNT" != "true" ]; then
if [ -n "$LLAMA_STACK_DIR" ]; then
CLI_ARGS+=("-v" "$(readlink -f "$LLAMA_STACK_DIR"):$stack_mount")
fi
if [ -n "$LLAMA_STACK_CLIENT_DIR" ]; then
CLI_ARGS+=("-v" "$(readlink -f "$LLAMA_STACK_CLIENT_DIR"):$client_mount")
fi
fi
if is_command_available selinuxenabled && selinuxenabled; then
# Disable SELinux labels -- we don't want to relabel the llama-stack source dir
CLI_ARGS+=("--security-opt" "label=disable")
fi
# Set version tag based on PyPI version
if [ -n "$PYPI_VERSION" ]; then
version_tag="$PYPI_VERSION"
elif [ -n "$TEST_PYPI_VERSION" ]; then
version_tag="test-$TEST_PYPI_VERSION"
elif [[ -n "$LLAMA_STACK_DIR" || -n "$LLAMA_STACK_CLIENT_DIR" ]]; then
version_tag="dev"
else
URL="https://pypi.org/pypi/llama-stack/json"
version_tag=$(curl -s $URL | jq -r '.info.version')
fi
# Add version tag to image name
image_tag="$image_name:$version_tag"
# Detect platform architecture
ARCH=$(uname -m)
if [ -n "$BUILD_PLATFORM" ]; then
CLI_ARGS+=("--platform" "$BUILD_PLATFORM")
elif [ "$ARCH" = "arm64" ] || [ "$ARCH" = "aarch64" ]; then
CLI_ARGS+=("--platform" "linux/arm64")
elif [ "$ARCH" = "x86_64" ]; then
CLI_ARGS+=("--platform" "linux/amd64")
else
echo "Unsupported architecture: $ARCH"
exit 1
fi
echo "PWD: $(pwd)"
echo "Containerfile: $TEMP_DIR/Containerfile"
set -x
$CONTAINER_BINARY build \
"${CLI_ARGS[@]}" \
-t "$image_tag" \
-f "$TEMP_DIR/Containerfile" \
"$BUILD_CONTEXT_DIR"
# clean up tmp/configs
rm -rf "$BUILD_CONTEXT_DIR/run.yaml" "$TEMP_DIR"
set +x
echo "Success!"

220
llama_stack/core/build_venv.sh
View file

@ -1,220 +0,0 @@
#!/bin/bash
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
LLAMA_STACK_DIR=${LLAMA_STACK_DIR:-}
LLAMA_STACK_CLIENT_DIR=${LLAMA_STACK_CLIENT_DIR:-}
TEST_PYPI_VERSION=${TEST_PYPI_VERSION:-}
# This timeout (in seconds) is necessary when installing PyTorch via uv since it's likely to time out
# Reference: https://github.com/astral-sh/uv/pull/1694
UV_HTTP_TIMEOUT=${UV_HTTP_TIMEOUT:-500}
UV_SYSTEM_PYTHON=${UV_SYSTEM_PYTHON:-}
VIRTUAL_ENV=${VIRTUAL_ENV:-}
set -euo pipefail
# Define color codes
RED='\033[0;31m'
NC='\033[0m' # No Color
SCRIPT_DIR=$(dirname "$(readlink -f "$0")")
source "$SCRIPT_DIR/common.sh"
# Usage function
usage() {
echo "Usage: $0 --env-name <env_name> --normal-deps <pip_dependencies> [--external-provider-deps <external_provider_deps>] [--optional-deps <special_pip_deps>]"
echo "Example: $0 --env-name mybuild --normal-deps 'numpy pandas scipy' --external-provider-deps 'foo' --optional-deps 'bar'"
exit 1
}
# Parse arguments
env_name=""
normal_deps=""
external_provider_deps=""
optional_deps=""
while [[ $# -gt 0 ]]; do
key="$1"
case "$key" in
--env-name)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --env-name requires a string value" >&2
usage
fi
env_name="$2"
shift 2
;;
--normal-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --normal-deps requires a string value" >&2
usage
fi
normal_deps="$2"
shift 2
;;
--external-provider-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --external-provider-deps requires a string value" >&2
usage
fi
external_provider_deps="$2"
shift 2
;;
--optional-deps)
if [[ -z "$2" || "$2" == --* ]]; then
echo "Error: --optional-deps requires a string value" >&2
usage
fi
optional_deps="$2"
shift 2
;;
*)
echo "Unknown option: $1" >&2
usage
;;
esac
done
# Check required arguments
if [[ -z "$env_name" || -z "$normal_deps" ]]; then
echo "Error: --env-name and --normal-deps are required." >&2
usage
fi
if [ -n "$LLAMA_STACK_DIR" ]; then
echo "Using llama-stack-dir=$LLAMA_STACK_DIR"
fi
if [ -n "$LLAMA_STACK_CLIENT_DIR" ]; then
echo "Using llama-stack-client-dir=$LLAMA_STACK_CLIENT_DIR"
fi
ENVNAME=""
# pre-run checks to make sure we can proceed with the installation
pre_run_checks() {
local env_name="$1"
if ! is_command_available uv; then
echo "uv is not installed, trying to install it."
if ! is_command_available pip; then
echo "pip is not installed, cannot automatically install 'uv'."
echo "Follow this link to install it:"
echo "https://docs.astral.sh/uv/getting-started/installation/"
exit 1
else
pip install uv
fi
fi
# checking if an environment with the same name already exists
if [ -d "$env_name" ]; then
echo "Environment '$env_name' already exists, re-using it."
fi
}
run() {
# Use only global variables set by flag parser
if [ -n "$UV_SYSTEM_PYTHON" ] || [ "$env_name" == "__system__" ]; then
echo "Installing dependencies in system Python environment"
export UV_SYSTEM_PYTHON=1
elif [ "$VIRTUAL_ENV" == "$env_name" ]; then
echo "Virtual environment $env_name is already active"
else
echo "Using virtual environment $env_name"
uv venv "$env_name"
source "$env_name/bin/activate"
fi
if [ -n "$TEST_PYPI_VERSION" ]; then
uv pip install fastapi libcst
uv pip install --extra-index-url https://test.pypi.org/simple/ \
--index-strategy unsafe-best-match \
llama-stack=="$TEST_PYPI_VERSION" \
$normal_deps
if [ -n "$optional_deps" ]; then
IFS='#' read -ra parts <<<"$optional_deps"
for part in "${parts[@]}"; do
echo "$part"
uv pip install $part
done
fi
if [ -n "$external_provider_deps" ]; then
IFS='#' read -ra parts <<<"$external_provider_deps"
for part in "${parts[@]}"; do
echo "$part"
uv pip install "$part"
done
fi
else
if [ -n "$LLAMA_STACK_DIR" ]; then
# only warn if DIR does not start with "git+"
if [ ! -d "$LLAMA_STACK_DIR" ] && [[ "$LLAMA_STACK_DIR" != git+* ]]; then
printf "${RED}Warning: LLAMA_STACK_DIR is set but directory does not exist: %s${NC}\n" "$LLAMA_STACK_DIR" >&2
exit 1
fi
printf "Installing from LLAMA_STACK_DIR: %s\n" "$LLAMA_STACK_DIR"
# editable only if LLAMA_STACK_DIR does not start with "git+"
if [[ "$LLAMA_STACK_DIR" != git+* ]]; then
EDITABLE="-e"
else
EDITABLE=""
fi
uv pip install --no-cache-dir $EDITABLE "$LLAMA_STACK_DIR"
else
uv pip install --no-cache-dir llama-stack
fi
if [ -n "$LLAMA_STACK_CLIENT_DIR" ]; then
# only warn if DIR does not start with "git+"
if [ ! -d "$LLAMA_STACK_CLIENT_DIR" ] && [[ "$LLAMA_STACK_CLIENT_DIR" != git+* ]]; then
printf "${RED}Warning: LLAMA_STACK_CLIENT_DIR is set but directory does not exist: %s${NC}\n" "$LLAMA_STACK_CLIENT_DIR" >&2
exit 1
fi
printf "Installing from LLAMA_STACK_CLIENT_DIR: %s\n" "$LLAMA_STACK_CLIENT_DIR"
# editable only if LLAMA_STACK_CLIENT_DIR does not start with "git+"
if [[ "$LLAMA_STACK_CLIENT_DIR" != git+* ]]; then
EDITABLE="-e"
else
EDITABLE=""
fi
uv pip install --no-cache-dir $EDITABLE "$LLAMA_STACK_CLIENT_DIR"
fi
printf "Installing pip dependencies\n"
uv pip install $normal_deps
if [ -n "$optional_deps" ]; then
IFS='#' read -ra parts <<<"$optional_deps"
for part in "${parts[@]}"; do
echo "Installing special provider module: $part"
uv pip install $part
done
fi
if [ -n "$external_provider_deps" ]; then
IFS='#' read -ra parts <<<"$external_provider_deps"
for part in "${parts[@]}"; do
echo "Installing external provider module: $part"
uv pip install "$part"
echo "Getting provider spec for module: $part and installing dependencies"
package_name=$(echo "$part" | sed 's/[<>=!].*//')
python3 -c "
import importlib
import sys
try:
module = importlib.import_module(f'$package_name.provider')
spec = module.get_provider_spec()
if hasattr(spec, 'pip_packages') and spec.pip_packages:
print('\\n'.join(spec.pip_packages))
except Exception as e:
print(f'Error getting provider spec for $package_name: {e}', file=sys.stderr)
" | uv pip install -r -
done
fi
fi
}
pre_run_checks "$env_name"
run

2
llama_stack/core/library_client.py
View file

@ -278,7 +278,7 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
else:
prefix = "!" if in_notebook() else ""
cprint(
f"Please run:\n\n{prefix}llama stack build --distro {self.config_path_or_distro_name} --image-type venv\n\n",
f"Please run:\n\n{prefix}llama stack list-deps {self.config_path_or_distro_name} | xargs -L1 uv pip install\n\n",
"yellow",
file=sys.stderr,
)

2
llama_stack/core/ui/README.md
View file

@ -9,7 +9,7 @@
1. Start up Llama Stack API server. More details [here](https://llamastack.github.io/latest/getting_started/index.htmll).
```
llama stack build --distro together --image-type venv
llama stack list-deps together | xargs -L1 uv pip install
llama stack run together
```

2
llama_stack/distributions/dell/doc_template.md
View file

@ -157,7 +157,7 @@ docker run \
Make sure you have done `pip install llama-stack` and have the Llama Stack CLI available.
```bash
llama stack build --distro {{ name }} --image-type conda
llama stack list-deps {{ name }} | xargs -L1 pip install
INFERENCE_MODEL=$INFERENCE_MODEL \
DEH_URL=$DEH_URL \
CHROMA_URL=$CHROMA_URL \

2
llama_stack/providers/remote/datasetio/nvidia/README.md
View file

@ -20,7 +20,7 @@ This provider enables dataset management using NVIDIA's NeMo Customizer service.
Build the NVIDIA environment:
```bash
llama stack build --distro nvidia --image-type venv
uv run llama stack list-deps nvidia | xargs -L1 uv pip install
```
### Basic Usage using the LlamaStack Python Client

2
llama_stack/providers/remote/inference/nvidia/NVIDIA.md
View file

@ -18,7 +18,7 @@ This provider enables running inference using NVIDIA NIM.
Build the NVIDIA environment:
```bash
llama stack build --distro nvidia --image-type venv
uv run llama stack list-deps nvidia | xargs -L1 uv pip install
```
### Basic Usage using the LlamaStack Python Client

2
llama_stack/providers/remote/inference/nvidia/__init__.py
View file

@ -10,7 +10,7 @@ from .config import NVIDIAConfig
async def get_adapter_impl(config: NVIDIAConfig, _deps) -> Inference:
# import dynamically so `llama stack build` does not fail due to missing dependencies
# import dynamically so `llama stack list-deps` does not fail due to missing dependencies
from .nvidia import NVIDIAInferenceAdapter
if not isinstance(config, NVIDIAConfig):

2
llama_stack/providers/remote/post_training/nvidia/README.md
View file

@ -22,7 +22,7 @@ This provider enables fine-tuning of LLMs using NVIDIA's NeMo Customizer service
Build the NVIDIA environment:
```bash
llama stack build --distro nvidia --image-type venv
uv run llama stack list-deps nvidia | xargs -L1 uv pip install
```
### Basic Usage using the LlamaStack Python Client

2
llama_stack/providers/remote/safety/nvidia/README.md
View file

@ -19,7 +19,7 @@ This provider enables safety checks and guardrails for LLM interactions using NV
Build the NVIDIA environment:
```bash
llama stack build --distro nvidia --image-type venv
uv run llama stack list-deps nvidia | xargs -L1 uv pip install
```
### Basic Usage using the LlamaStack Python Client

2
tests/integration/fixtures/common.py
View file

@ -40,7 +40,7 @@ def is_port_available(port: int, host: str = "localhost") -> bool:
def start_llama_stack_server(config_name: str) -> subprocess.Popen:
"""Start a llama stack server with the given config."""
cmd = f"uv run --with llama-stack llama stack build --distro {config_name} --image-type venv --run"
cmd = f"uv run llama stack run {config_name}"
devnull = open(os.devnull, "w")
process = subprocess.Popen(
shlex.split(cmd),

40
tests/unit/distribution/test_build_path.py
View file

@ -1,40 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from pathlib import Path
from llama_stack.cli.stack._build import (
_run_stack_build_command_from_build_config,
)
from llama_stack.core.datatypes import BuildConfig, DistributionSpec
from llama_stack.core.utils.image_types import LlamaStackImageType
def test_container_build_passes_path(monkeypatch, tmp_path):
called_with = {}
def spy_build_image(build_config, image_name, distro_or_config, run_config=None):
called_with["path"] = distro_or_config
called_with["run_config"] = run_config
return 0
monkeypatch.setattr(
"llama_stack.cli.stack._build.build_image",
spy_build_image,
raising=True,
)
cfg = BuildConfig(
image_type=LlamaStackImageType.CONTAINER.value,
distribution_spec=DistributionSpec(providers={}, description=""),
)
_run_stack_build_command_from_build_config(cfg, image_name="dummy")
assert "path" in called_with
assert isinstance(called_with["path"], str)
assert Path(called_with["path"]).exists()
assert called_with["run_config"] is None