Update main post-23.10 release (#6484)

* Update README and versions for 23.10 branch (#6399)

* Cherry-picking vLLM backend changes (#6404)

* Update build.py to build vLLM backend (#6394)

* Add Python backend when vLLM backend built (#6397)

---------

Co-authored-by: dyastremsky <[email protected]>

* Add documentation on request cancellation (#6403) (#6407)

* Add documentation on request cancellation

* Include python backend

* Update docs/user_guide/request_cancellation.md

* Update docs/user_guide/request_cancellation.md

* Update docs/README.md

* Update docs/user_guide/request_cancellation.md

* Remove inflight term from the main documentation

* Address review comments

* Fix

* Update docs/user_guide/request_cancellation.md

* Fix

---------

Co-authored-by: Iman Tabrizian <[email protected]>
Co-authored-by: Neelay Shah <[email protected]>
Co-authored-by: Ryan McCormick <[email protected]>
Co-authored-by: Jacky <[email protected]>

* Fixes in request cancellation doc (#6409) (#6410)

* TRT-LLM backend build changes (#6406) (#6430)

* Update url

* Debugging

* Debugging

* Update url

* Fix build for TRT-LLM backend

* Remove TRTLLM TRT and CUDA versions

* Fix up unused var

* Fix up dir name

* FIx cmake patch

* Remove previous TRT version

* Install required packages for example models

* Remove packages that are only needed for testing

* Fixing vllm build (#6433) (#6437)

* Fixing torch version for vllm

Co-authored-by: Olga Andreeva <[email protected]>

* Update TRT-LLM backend url (#6455) (#6460)

* TRTLLM backend post release

* TRTLLM backend post release

* Update submodule url for permission issue

* Update submodule url

* Fix up

* Not using postbuild function to workaround submodule url permission issue

* remove redundant lines

* Revert "remove redundant lines"

This reverts commit 86be7ad.

* restore missed lines

* Update build.py

Co-authored-by: Olga Andreeva <[email protected]>

* Update build.py

Co-authored-by: Olga Andreeva <[email protected]>

---------

Co-authored-by: Tanmay Verma <[email protected]>
Co-authored-by: dyastremsky <[email protected]>
Co-authored-by: Iman Tabrizian <[email protected]>
Co-authored-by: Neelay Shah <[email protected]>
Co-authored-by: Ryan McCormick <[email protected]>
Co-authored-by: Jacky <[email protected]>
Co-authored-by: Kris Hung <[email protected]>
Co-authored-by: Katherine Yang <[email protected]>
Co-authored-by: Olga Andreeva <[email protected]>

Author: 10 people

Dockerfile.sdk

@@ -29,7 +29,7 @@
 #
 
 # Base image on the minimum Triton container
-ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:23.09-py3-min
+ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:23.10-py3-min
 
 ARG TRITON_CLIENT_REPO_SUBDIR=clientrepo
 ARG TRITON_COMMON_REPO_TAG=main

README.md

@@ -93,16 +93,16 @@ Inference Server with the
 
 ```bash
 # Step 1: Create the example model repository
-git clone -b r23.09 https://github.com/triton-inference-server/server.git
+git clone -b r23.10 https://github.com/triton-inference-server/server.git
 cd server/docs/examples
 ./fetch_models.sh
 
 # Step 2: Launch triton from the NGC Triton container
-docker run --gpus=1 --rm --net=host -v ${PWD}/model_repository:/models nvcr.io/nvidia/tritonserver:23.09-py3 tritonserver --model-repository=/models
+docker run --gpus=1 --rm --net=host -v ${PWD}/model_repository:/models nvcr.io/nvidia/tritonserver:23.10-py3 tritonserver --model-repository=/models
 
 # Step 3: Sending an Inference Request
 # In a separate console, launch the image_client example from the NGC Triton SDK container
-docker run -it --rm --net=host nvcr.io/nvidia/tritonserver:23.09-py3-sdk
+docker run -it --rm --net=host nvcr.io/nvidia/tritonserver:23.10-py3-sdk
 /workspace/install/bin/image_client -m densenet_onnx -c 3 -s INCEPTION /workspace/images/mug.jpg
 
 # Inference should return the following

build.py

@@ -72,7 +72,7 @@
 TRITON_VERSION_MAP = {
     "2.40.0dev": (
         "23.11dev",  # triton container
-        "23.09",  # upstream container
+        "23.10",  # upstream container
         "1.16.0",  # ORT
         "2023.0.0",  # ORT OpenVINO
         "2023.0.0",  # Standalone OpenVINO

deploy/aws/values.yaml

@@ -27,7 +27,7 @@
 replicaCount: 1
 
 image:
-  imageName: nvcr.io/nvidia/tritonserver:23.09-py3
+  imageName: nvcr.io/nvidia/tritonserver:23.10-py3
   pullPolicy: IfNotPresent
   modelRepositoryPath: s3://triton-inference-server-repository/model_repository
   numGpus: 1

deploy/fleetcommand/Chart.yaml

@@ -26,7 +26,7 @@
 
 apiVersion: v1
 # appVersion is the Triton version; update when changing release
-appVersion: "2.38.0"
+appVersion: "2.39.0"
 description: Triton Inference Server (Fleet Command)
 name: triton-inference-server
 # version is the Chart version; update when changing anything in the chart

deploy/fleetcommand/values.yaml

@@ -27,7 +27,7 @@
 replicaCount: 1
 
 image:
-  imageName: nvcr.io/nvidia/tritonserver:23.09-py3
+  imageName: nvcr.io/nvidia/tritonserver:23.10-py3
   pullPolicy: IfNotPresent
   numGpus: 1
   serverCommand: tritonserver
@@ -46,13 +46,13 @@ image:
     # Model Control Mode (Optional, default: none)
     #
     # To set model control mode, uncomment and configure below
-    # See https://github.com/triton-inference-server/server/blob/r23.09/docs/model_management.md
+    # See https://github.com/triton-inference-server/server/blob/r23.10/docs/model_management.md
     #  for more details
     #- --model-control-mode=explicit|poll|none
     #
     # Additional server args
     #
-    # see https://github.com/triton-inference-server/server/blob/r23.09/README.md
+    # see https://github.com/triton-inference-server/server/blob/r23.10/README.md
     #  for more details
 
 service:

deploy/gcp/values.yaml

@@ -27,7 +27,7 @@
 replicaCount: 1
 
 image:
-  imageName: nvcr.io/nvidia/tritonserver:23.09-py3
+  imageName: nvcr.io/nvidia/tritonserver:23.10-py3
   pullPolicy: IfNotPresent
   modelRepositoryPath: gs://triton-inference-server-repository/model_repository
   numGpus: 1

deploy/gke-marketplace-app/benchmark/perf-analyzer-script/triton_client.yaml

@@ -33,7 +33,7 @@ metadata:
   namespace: default
 spec:
   containers:
-  - image: nvcr.io/nvidia/tritonserver:23.09-py3-sdk
+  - image: nvcr.io/nvidia/tritonserver:23.10-py3-sdk
     imagePullPolicy: Always
     name: nv-triton-client
     securityContext:

deploy/gke-marketplace-app/server-deployer/build_and_push.sh

@@ -27,9 +27,9 @@
 
 export REGISTRY=gcr.io/$(gcloud config get-value project | tr ':' '/')
 export APP_NAME=tritonserver
-export MAJOR_VERSION=2.38
-export MINOR_VERSION=2.38.0
-export NGC_VERSION=23.09-py3
+export MAJOR_VERSION=2.39
+export MINOR_VERSION=2.39.0
+export NGC_VERSION=23.10-py3
 
 docker pull nvcr.io/nvidia/$APP_NAME:$NGC_VERSION
 

deploy/gke-marketplace-app/server-deployer/chart/triton/Chart.yaml

@@ -25,7 +25,7 @@
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 apiVersion: v1
-appVersion: "2.38"
+appVersion: "2.39"
 description: Triton Inference Server
 name: triton-inference-server
-version: 2.38.0
+version: 2.39.0

deploy/gke-marketplace-app/server-deployer/chart/triton/values.yaml

@@ -32,13 +32,13 @@ tritonProtocol: HTTP
 # HPA GPU utilization autoscaling target
 HPATargetAverageValue: 85
 modelRepositoryPath: gs://triton_sample_models/23_09
-publishedVersion: '2.38.0'
+publishedVersion: '2.39.0'
 gcpMarketplace: true
 
 image:
   registry: gcr.io
   repository: nvidia-ngc-public/tritonserver
-  tag: 23.09-py3
+  tag: 23.10-py3
   pullPolicy: IfNotPresent
   # modify the model repository here to match your GCP storage bucket
   numGpus: 1

deploy/gke-marketplace-app/server-deployer/data-test/schema.yaml

@@ -27,7 +27,7 @@
 x-google-marketplace:
   schemaVersion: v2
   applicationApiVersion: v1beta1
-  publishedVersion: '2.38.0'
+  publishedVersion: '2.39.0'
   publishedVersionMetadata:
     releaseNote: >-
       Initial release.

deploy/gke-marketplace-app/server-deployer/schema.yaml

@@ -27,7 +27,7 @@
 x-google-marketplace:
   schemaVersion: v2
   applicationApiVersion: v1beta1
-  publishedVersion: '2.38.0'
+  publishedVersion: '2.39.0'
   publishedVersionMetadata:
     releaseNote: >-
       Initial release.

deploy/gke-marketplace-app/trt-engine/README.md

@@ -33,7 +33,7 @@
 ```
 docker run --gpus all -it --network host \
     --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 \
-    -v ~:/scripts nvcr.io/nvidia/tensorrt:23.09-py3
+    -v ~:/scripts nvcr.io/nvidia/tensorrt:23.10-py3
 
 pip install onnx six torch tf2onnx tensorflow
 

deploy/k8s-onprem/values.yaml

@@ -29,7 +29,7 @@ tags:
   loadBalancing: true
 
 image:
-  imageName: nvcr.io/nvidia/tritonserver:23.09-py3
+  imageName: nvcr.io/nvidia/tritonserver:23.10-py3
   pullPolicy: IfNotPresent
   modelRepositoryServer: < Replace with the IP Address of your file server >
   modelRepositoryPath: /srv/models

docs/customization_guide/build.md

@@ -173,7 +173,7 @@ $ ./build.py ... --repo-tag=common:<container tag> --repo-tag=core:<container ta
 
 If you are building on a release branch then `<container tag>` will
 default to the branch name. For example, if you are building on the
-r23.09 branch, `<container tag>` will default to r23.09. If you are
+r23.10 branch, `<container tag>` will default to r23.10. If you are
 building on any other branch (including the *main* branch) then
 `<container tag>` will default to "main". Therefore, you typically do
 not need to provide `<container tag>` at all (nor the preceding
@@ -334,8 +334,8 @@ python build.py --cmake-dir=<path/to/repo>/build --build-dir=/tmp/citritonbuild
 If you are building on *main* branch then '<container tag>' will
 default to "main". If you are building on a release branch then
 '<container tag>' will default to the branch name. For example, if you
-are building on the r23.09 branch, '<container tag>' will default to
-r23.09. Therefore, you typically do not need to provide '<container
+are building on the r23.10 branch, '<container tag>' will default to
+r23.10. Therefore, you typically do not need to provide '<container
 tag>' at all (nor the preceding colon). You can use a different
 '<container tag>' for a component to instead use the corresponding
 branch/tag in the build. For example, if you have a branch called

docs/customization_guide/compose.md

@@ -44,8 +44,8 @@ from source to get more exact customization.
 The `compose.py` script can be found in the [server repository](https://github.com/triton-inference-server/server).
 Simply clone the repository and run `compose.py` to create a custom container.
 Note: Created container version will depend on the branch that was cloned.
-For example branch [r23.09](https://github.com/triton-inference-server/server/tree/r23.09)
-should be used to create a image based on the NGC 23.09 Triton release.
+For example branch [r23.10](https://github.com/triton-inference-server/server/tree/r23.10)
+should be used to create a image based on the NGC 23.10 Triton release.
 
 `compose.py` provides `--backend`, `--repoagent` options that allow you to
 specify which backends and repository agents to include in the custom image.
@@ -76,19 +76,19 @@ For example, running
 ```
 python3 compose.py --backend tensorflow1 --repoagent checksum
 ```
-on branch [r23.09](https://github.com/triton-inference-server/server/tree/r23.09) pulls:
-- `min` container `nvcr.io/nvidia/tritonserver:23.09-py3-min`
-- `full` container `nvcr.io/nvidia/tritonserver:23.09-py3`
+on branch [r23.10](https://github.com/triton-inference-server/server/tree/r23.10) pulls:
+- `min` container `nvcr.io/nvidia/tritonserver:23.10-py3-min`
+- `full` container `nvcr.io/nvidia/tritonserver:23.10-py3`
 
 Alternatively, users can specify the version of Triton container to pull from any branch by either:
 1. Adding flag `--container-version <container version>` to branch
 ```
-python3 compose.py --backend tensorflow1 --repoagent checksum --container-version 23.09
+python3 compose.py --backend tensorflow1 --repoagent checksum --container-version 23.10
 ```
 2. Specifying `--image min,<min container image name> --image full,<full container image name>`.
    The user is responsible for specifying compatible `min` and `full` containers.
 ```
-python3 compose.py --backend tensorflow1 --repoagent checksum --image min,nvcr.io/nvidia/tritonserver:23.09-py3-min --image full,nvcr.io/nvidia/tritonserver:23.09-py3
+python3 compose.py --backend tensorflow1 --repoagent checksum --image min,nvcr.io/nvidia/tritonserver:23.10-py3-min --image full,nvcr.io/nvidia/tritonserver:23.10-py3
 ```
 Method 1 and 2 will result in the same composed container. Furthermore, `--image` flag overrides the `--container-version` flag when both are specified.
 

docs/customization_guide/test.md

@@ -49,7 +49,7 @@ $ ./gen_qa_custom_ops
 ```
 
 This will create multiple model repositories in /tmp/<version>/qa_*
-(for example /tmp/23.09/qa_model_repository).  The TensorRT models
+(for example /tmp/23.10/qa_model_repository).  The TensorRT models
 will be created for the GPU on the system that CUDA considers device 0
 (zero). If you have multiple GPUs on your system see the documentation
 in the scripts for how to target a specific GPU.

docs/user_guide/custom_operations.md

@@ -64,7 +64,7 @@ simple way to ensure you are using the correct version of TensorRT is
 to use the [NGC TensorRT
 container](https://ngc.nvidia.com/catalog/containers/nvidia:tensorrt)
 corresponding to the Triton container. For example, if you are using
-the 23.09 version of Triton, use the 23.09 version of the TensorRT
+the 23.10 version of Triton, use the 23.10 version of the TensorRT
 container.
 
 ## TensorFlow
@@ -123,7 +123,7 @@ simple way to ensure you are using the correct version of TensorFlow
 is to use the [NGC TensorFlow
 container](https://ngc.nvidia.com/catalog/containers/nvidia:tensorflow)
 corresponding to the Triton container. For example, if you are using
-the 23.09 version of Triton, use the 23.09 version of the TensorFlow
+the 23.10 version of Triton, use the 23.10 version of the TensorFlow
 container.
 
 ## PyTorch
@@ -167,7 +167,7 @@ simple way to ensure you are using the correct version of PyTorch is
 to use the [NGC PyTorch
 container](https://ngc.nvidia.com/catalog/containers/nvidia:pytorch)
 corresponding to the Triton container. For example, if you are using
-the 23.09 version of Triton, use the 23.09 version of the PyTorch
+the 23.10 version of Triton, use the 23.10 version of the PyTorch
 container.
 
 ## ONNX

docs/user_guide/performance_tuning.md

@@ -235,7 +235,7 @@ with a `tritonserver` binary.
 
 ```bash
 # Start server container
-docker run -ti --rm --gpus=all --network=host -v $PWD:/mnt --name triton-server nvcr.io/nvidia/tritonserver:23.09-py3
+docker run -ti --rm --gpus=all --network=host -v $PWD:/mnt --name triton-server nvcr.io/nvidia/tritonserver:23.10-py3
 
 # Start serving your models
 tritonserver --model-repository=/mnt/models
@@ -284,7 +284,7 @@ by setting the `-u` flag, such as `perf_analyzer -m densenet_onnx -u
 
 ```bash
 # Start the SDK container interactively
-docker run -ti --rm --gpus=all --network=host -v $PWD:/mnt --name triton-client nvcr.io/nvidia/tritonserver:23.09-py3-sdk
+docker run -ti --rm --gpus=all --network=host -v $PWD:/mnt --name triton-client nvcr.io/nvidia/tritonserver:23.10-py3-sdk
 
 # Benchmark model being served from step 3
 perf_analyzer -m densenet_onnx --concurrency-range 1:4

qa/common/gen_jetson_trt_models

@@ -33,9 +33,8 @@
 #!/bin/bash -xe
 # Make all generated files accessible outside of container
 umask 0000
-
 # Set the version of the models
-TRITON_VERSION=${TRITON_VERSION:=23.09}
+TRITON_VERSION=${TRITON_VERSION:=23.10}
 # Set the CUDA device to use
 CUDA_DEVICE=${RUNNER_ID:=0}
 # Set TensorRT image

qa/common/gen_qa_custom_ops

@@ -37,7 +37,7 @@
 ##
 ############################################################################
 
-TRITON_VERSION=${TRITON_VERSION:=23.09}
+TRITON_VERSION=${TRITON_VERSION:=23.10}
 NVIDIA_UPSTREAM_VERSION=${NVIDIA_UPSTREAM_VERSION:=$TRITON_VERSION}
 TENSORFLOW_IMAGE=${TENSORFLOW_IMAGE:=nvcr.io/nvidia/tensorflow:$NVIDIA_UPSTREAM_VERSION-tf2-py3}
 PYTORCH_IMAGE=${PYTORCH_IMAGE:=nvcr.io/nvidia/pytorch:$NVIDIA_UPSTREAM_VERSION-py3}

qa/common/gen_qa_model_repository

@@ -48,7 +48,7 @@
 ##
 ############################################################################
 
-TRITON_VERSION=${TRITON_VERSION:=23.09}
+TRITON_VERSION=${TRITON_VERSION:=23.10}
 
 # ONNX. Use ONNX_OPSET 0 to use the default for ONNX version
 ONNX_VERSION=1.13.0