agg_router.sh 1.03 KB
Newer Older
1
#!/bin/bash
2
# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3
4
# SPDX-License-Identifier: Apache-2.0

5
6
7
8
9
10
set -e
trap 'echo Cleaning up...; kill 0' EXIT

SCRIPT_DIR="$(dirname "$(readlink -f "$0")")"
source "$SCRIPT_DIR/../../../common/launch_utils.sh"

11
# Environment variables with defaults
12
export DYNAMO_HOME=${DYNAMO_HOME:-"/workspace"}
13
14
export MODEL_PATH=${MODEL_PATH:-"Qwen/Qwen3-0.6B"}
export SERVED_MODEL_NAME=${SERVED_MODEL_NAME:-"Qwen/Qwen3-0.6B"}
15
export AGG_ENGINE_ARGS=${AGG_ENGINE_ARGS:-"$DYNAMO_HOME/examples/backends/trtllm/engine_configs/qwen3/agg.yaml"}
16

17
# run frontend
18
19
# dynamo.frontend accepts either --http-port flag or DYN_HTTP_PORT env var (defaults to 8000)
python3 -m dynamo.frontend --router-mode kv &
20
21

# run worker
22
python3 -m dynamo.trtllm \
23
24
25
  --model-path "$MODEL_PATH" \
  --served-model-name "$SERVED_MODEL_NAME" \
  --extra-engine-args "$AGG_ENGINE_ARGS" \
26
27
28
29
  --publish-events-and-metrics &

# Exit on first worker failure; kill 0 in the EXIT trap tears down the rest
wait_any_exit