Merge pull request #1150 from basetenlabs/bump-version-0.9.36

Release 0.9.36

.devcontainer/Dockerfile

@@ -21,7 +21,7 @@ RUN apt-get update && /bin/bash /tmp/library-scripts/common-debian.sh "${INSTALL
 
 USER vscode
 RUN curl -sSL https://install.python-poetry.org | python
-ENV PATH=/home/vscode/.poetry/bin:$PATH
+ENV PATH="/home/vscode/.poetry/bin:$PATH"
 
 USER root
 

.devcontainer/gpu/Dockerfile

@@ -21,7 +21,7 @@ RUN apt-get update && /bin/bash /tmp/library-scripts/common-debian.sh "${INSTALL
 
 USER vscode
 RUN curl -sSL https://install.python-poetry.org | python
-ENV PATH=/home/vscode/.poetry/bin:$PATH
+ENV PATH="/home/vscode/.poetry/bin:$PATH"
 
 USER root
 

.github/workflows/pr.yml

@@ -33,7 +33,7 @@ jobs:
       - uses: ./.github/actions/setup-python/
       - run: poetry install
       - name: run tests
-        run: poetry run pytest -v --cov=truss -m 'not integration' --junitxml=report.xml
+        run: poetry run pytest --durations=0  -m 'not integration' --junitxml=report.xml
       - name: Publish Test Report  # Not sure how to display this in the UI for non PRs.
         uses: mikepenz/action-junit-report@v4
         if: always()

docker/base_images/base_image.Dockerfile.jinja

@@ -1,8 +1,8 @@
 {% if use_gpu %}
 FROM nvidia/cuda:12.2.2-base-ubuntu20.04
-ENV CUDNN_VERSION=8.9.5.29
-ENV CUDA=12.2
-ENV LD_LIBRARY_PATH /usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH
+ENV CUDNN_VERSION="8.9.5.29"
+ENV CUDA="12.2"
+ENV LD_LIBRARY_PATH="/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH"
 
 RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub && \
     apt-get update && apt-get install -y --no-install-recommends \
@@ -21,8 +21,8 @@ RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/
     rm -rf /var/lib/apt/lists/*
 
 # Allow statements and log messages to immediately appear in the Knative logs
-ENV PYTHONUNBUFFERED True
-ENV DEBIAN_FRONTEND=noninteractive
+ENV PYTHONUNBUFFERED="True"
+ENV DEBIAN_FRONTEND="noninteractive"
 
 RUN apt update && \
     apt install -y bash \
@@ -49,7 +49,7 @@ FROM python:{{python_version}}
 RUN apt update && apt install -y
 
 # Allow statements and log messages to immediately appear in the Knative logs
-ENV PYTHONUNBUFFERED True
+ENV PYTHONUNBUFFERED="True"
 {% endif %}
 
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "truss"
-version = "0.9.35"
+version = "0.9.36"
 description = "A seamless bridge from model development to model delivery"
 license = "MIT"
 readme = "README.md"

truss-chains/examples/audio-transcription/README.md

@@ -14,15 +14,15 @@ whisper transcription model (which has slower deployment times), they are in
 the current setup deployed separately:
 
 ```bash
-truss chains deploy whisper_chainlet.py
+truss chains push whisper_chainlet.py
 ```
 
-Insert the predict URL for the Whisper Chainlet (printed by above deploy
+Insert the predict URL for the Whisper Chainlet (printed by above push
 command or can be found on the status page) as a value for
-`WHISPER_PREDICT_URL` in `transcribe.py`. The deploy the transcribe chain.
+`WHISPER_PREDICT_URL` in `transcribe.py`. Then push the transcribe chain.
 
 ```bash
-truss chains deploy transcribe.py
+truss chains push transcribe.py
 ```
 
 An example local invocation of the chain is given in the main-section of

truss-chains/examples/rag/README.md

@@ -47,7 +47,7 @@ python rag_chain.py
 Deploy the Chain to production:
 
 ```sh
-truss chains deploy rag_chain.py
+truss chains push rag_chain.py
 ```
 
 Note that this command will print you with an example cURL command how to
@@ -57,7 +57,7 @@ For example a chain invocation might look like this (you need to update the
 URL):
 
 ```sh
-curl -X POST 'https://model-5wo86nn3.api.baseten.co/development/predict' \
+curl -X POST 'https://chain-<CHAIN_ID>.api.baseten.co/development/run_remote' \
     -H "Authorization: Api-Key $BASETEN_API_KEY" \
     -d '{"new_bio": "Sam just moved to Manhattan for his new job at a large bank.In college, he enjoyed building sets for student plays."}'
 ```

truss-chains/truss_chains/remote.py

@@ -132,9 +132,8 @@ def _push_service(
     else:
         raise NotImplementedError(options)
 
-    logging.info(
-        f"Pushed `{chainlet_descriptor.display_name}` @ {service.predict_url}."
-    )
+    logging.info(f"Pushed `{chainlet_descriptor.display_name}`")
+    logging.debug(f"Internal model endpoint: `{service.predict_url}`")
     return service
 
 

truss/__init__.py

@@ -1,7 +1,13 @@
+import warnings
 from pathlib import Path
 
+from pydantic import PydanticDeprecatedSince20
 from single_source import get_version
 
+# Suppress Pydantic V1 warnings, because we have to use it for backwards compat.
+warnings.filterwarnings("ignore", category=PydanticDeprecatedSince20)
+
+
 __version__ = get_version(__name__, Path(__file__).parent.parent)
 
 

truss/cli/cli.py

@@ -470,7 +470,7 @@ def _create_chains_table(service) -> Tuple[rich.table.Table, List[str]]:
 @click.option(
     "--publish/--no-publish",
     type=bool,
-    default=True,
+    default=False,
     help="Create chainlets as published deployments.",
 )
 @click.option(

truss/config/trt_llm.py

@@ -1,11 +1,17 @@
 import json
 import logging
+import warnings
 from enum import Enum
 from typing import Optional
 
-from pydantic import BaseModel, validator
+from huggingface_hub.errors import HFValidationError
+from huggingface_hub.utils import validate_repo_id
+from pydantic import BaseModel, PydanticDeprecatedSince20, validator
 from rich.console import Console
 
+# Suppress Pydantic V1 warnings, because we have to use it for backwards compat.
+warnings.filterwarnings("ignore", category=PydanticDeprecatedSince20)
+
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
@@ -97,6 +103,8 @@ def __init__(self, **data):
         super().__init__(**data)
         self._validate_minimum_required_configuration()
         self._validate_kv_cache_flags()
+        if self.build.checkpoint_repository.source == CheckpointSource.HF:
+            self._validate_hf_repo_id()
 
     # In pydantic v2 this would be `@model_validator(mode="after")` and
     # the __init__ override can be removed.
@@ -131,6 +139,14 @@ def _validate_kv_cache_flags(self):
             raise ValueError("Using fp8 context fmha requires paged context fmha")
         return self
 
+    def _validate_hf_repo_id(self):
+        try:
+            validate_repo_id(self.build.checkpoint_repository.repo)
+        except HFValidationError as e:
+            raise ValueError(
+                f"HuggingFace repository validation failed: {str(e)}"
+            ) from e
+
     @property
     def requires_build(self):
         if self.build is not None:

truss/constants.py

@@ -103,7 +103,7 @@
 
 REGISTRY_BUILD_SECRET_PREFIX = "DOCKER_REGISTRY_"
 
-TRTLLM_BASE_IMAGE = "baseten/briton-server:5fa9436e_v0.0.9"
+TRTLLM_BASE_IMAGE = "baseten/briton-server:5fa9436e_v0.0.11"
 TRTLLM_PYTHON_EXECUTABLE = "/usr/bin/python3"
 BASE_TRTLLM_REQUIREMENTS = [
     "grpcio==1.62.3",

truss/contexts/image_builder/serving_image_builder.py

@@ -12,6 +12,7 @@
 from google.cloud import storage
 from huggingface_hub import get_hf_file_metadata, hf_hub_url, list_repo_files
 from huggingface_hub.utils import filter_repo_objects
+from truss import constants
 from truss.config.trt_llm import TrussTRTLLMModel
 from truss.constants import (
     AUDIO_MODEL_TRTLLM_REQUIREMENTS,
@@ -70,7 +71,6 @@
 GCS_CREDENTIALS = "service_account.json"
 S3_CREDENTIALS = "s3_credentials.json"
 
-HF_ACCESS_TOKEN_SECRET_NAME = "hf_access_token"
 HF_ACCESS_TOKEN_FILE_NAME = "hf-access-token"
 
 CLOUD_BUCKET_CACHE = Path("/app/model_cache/")
@@ -526,7 +526,7 @@ def _render_dockerfile(
             build_dir / USER_SUPPLIED_REQUIREMENTS_TXT_FILENAME
         )
 
-        hf_access_token = config.secrets.get(HF_ACCESS_TOKEN_SECRET_NAME)
+        hf_access_token = config.secrets.get(constants.HF_ACCESS_TOKEN_KEY)
         dockerfile_contents = dockerfile_template.render(
             should_install_server_requirements=should_install_server_requirements,
             base_image_name_and_tag=base_image_name_and_tag,

truss/remote/baseten/service.py

@@ -1,6 +1,7 @@
 import enum
 import time
 import urllib.parse
+import warnings
 from typing import (
     Any,
     Dict,
@@ -17,6 +18,9 @@
 from truss.truss_handle import TrussHandle
 from truss.util.errors import RemoteNetworkError
 
+# "classes created inside an enum will not become a member" -> intended here anyway.
+warnings.filterwarnings("ignore", category=DeprecationWarning, message=".*enum.*")
+
 DEFAULT_STREAM_ENCODING = "utf-8"
 
 

truss/remote/remote_factory.py

@@ -1,11 +1,12 @@
 import inspect
 
 try:
+    from configparser import DEFAULTSECT, ConfigParser  # type: ignore
+except ImportError:
+    # We need to do this for old python.
     from configparser import DEFAULTSECT
     from configparser import SafeConfigParser as ConfigParser
-except ImportError:
-    # We need to do this for py312 and onwards.
-    from configparser import DEFAULTSECT, ConfigParser  # type: ignore
+
 
 from functools import partial
 from operator import is_not

truss/templates/base.Dockerfile.jinja

@@ -1,7 +1,7 @@
 ARG PYVERSION={{config.python_version}}
-FROM {{base_image_name_and_tag}} as truss_server
+FROM {{base_image_name_and_tag}} AS truss_server
 
-ENV PYTHON_EXECUTABLE {{ config.base_image.python_executable_path or 'python3' }}
+ENV PYTHON_EXECUTABLE="{{ config.base_image.python_executable_path or 'python3' }}"
 
 {% block fail_fast %}
 RUN grep -w 'ID=debian\|ID_LIKE=debian' /etc/os-release || { echo "ERROR: Supplied base image is not a debian image"; exit 1; }
@@ -52,7 +52,7 @@ RUN pip install -r {{config_requirements_filename}} --no-cache-dir && rm -rf /ro
 
 
 
-ENV APP_HOME /app
+ENV APP_HOME="/app"
 WORKDIR $APP_HOME
 
 
@@ -68,7 +68,7 @@ COPY ./{{config.bundled_packages_dir}} /packages
 
 
 {% for env_var_name, env_var_value in config.environment_variables.items() %}
-ENV {{ env_var_name }} {{ env_var_value }}
+ENV {{ env_var_name }}="{{ env_var_value }}"
 {% endfor %}
 
 {% block run %}

truss/templates/cache.Dockerfile.jinja

@@ -1,14 +1,14 @@
-FROM python:3.11-slim as cache_warmer
+FROM python:3.11-slim AS cache_warmer
 
 RUN mkdir -p /app/model_cache
 WORKDIR /app
 
 {% if hf_access_token %}
-ENV HUGGING_FACE_HUB_TOKEN {{hf_access_token}}
+ENV HUGGING_FACE_HUB_TOKEN="{{hf_access_token}}"
 {% endif %}
 
 RUN apt-get -y update; apt-get -y install curl; curl -s https://baseten-public.s3.us-west-2.amazonaws.com/bin/b10cp-5fe8dc7da-linux-amd64 -o /app/b10cp; chmod +x /app/b10cp
-ENV B10CP_PATH_TRUSS /app/b10cp
+ENV B10CP_PATH_TRUSS="/app/b10cp"
 COPY ./cache_requirements.txt /app/cache_requirements.txt
 RUN pip install -r /app/cache_requirements.txt --no-cache-dir && rm -rf /root/.cache/pip
 COPY ./cache_warmer.py /cache_warmer.py

truss/templates/control/control/server.py

@@ -35,7 +35,7 @@ def run(self):
                 "inference_server_home": self._inf_serv_home,
                 "inference_server_process_args": [
                     self._python_executable_path,
-                    f"{self._inf_serv_home}/inference_server.py",
+                    f"{self._inf_serv_home}/main.py",
                 ],
                 "control_server_host": "0.0.0.0",
                 "control_server_port": self._control_server_port,

truss/templates/server.Dockerfile.jinja

@@ -7,8 +7,8 @@
 {% block base_image_patch %}
 # If user base image is supplied in config, apply build commands from truss base image
 {% if config.base_image %}
-ENV PYTHONUNBUFFERED True
-ENV DEBIAN_FRONTEND=noninteractive
+ENV PYTHONUNBUFFERED="True"
+ENV DEBIAN_FRONTEND="noninteractive"
 
 RUN apt update && \
     apt install -y bash \
@@ -90,14 +90,14 @@ COPY ./{{ config.model_module_dir }} /app/model
 
 {% block run %}
     {%- if config.live_reload %}
-ENV HASH_TRUSS {{truss_hash}}
-ENV CONTROL_SERVER_PORT 8080
-ENV INFERENCE_SERVER_PORT 8090
+ENV HASH_TRUSS="{{truss_hash}}"
+ENV CONTROL_SERVER_PORT="8080"
+ENV INFERENCE_SERVER_PORT="8090"
 ENV SERVER_START_CMD="/control/.env/bin/python3 /control/control/server.py"
 ENTRYPOINT ["/control/.env/bin/python3", "/control/control/server.py"]
     {%- else %}
-ENV INFERENCE_SERVER_PORT 8080
-ENV SERVER_START_CMD="{{(config.base_image.python_executable_path or "python3") ~ " /app/inference_server.py"}}"
-ENTRYPOINT ["{{config.base_image.python_executable_path or "python3"}}", "/app/inference_server.py"]
+ENV INFERENCE_SERVER_PORT="8080"
+ENV SERVER_START_CMD="{{(config.base_image.python_executable_path or "python3") ~ " /app/main.py"}}"
+ENTRYPOINT ["{{config.base_image.python_executable_path or "python3"}}", "/app/main.py"]
     {%- endif %}
 {% endblock %}