agg.sh 2.26 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:-"$DYNAMO_HOME/examples/backends/trtllm/engine_configs/qwen3/agg.yaml"}
16
17
18
export MODALITY=${MODALITY:-"text"}
# If you want to use multimodal, set MODALITY to "multimodal"
#export MODALITY=${MODALITY:-"multimodal"}
19
20


21
ENABLE_OTEL=false
22
EXTRA_ARGS=()
23
24
25
26
27
28
29
30
31
32
33
34
while [[ $# -gt 0 ]]; do
    case $1 in
        --enable-otel)
            ENABLE_OTEL=true
            shift
            ;;
        -h|--help)
            echo "Usage: $0 [OPTIONS]"
            echo "Options:"
            echo "  --enable-otel        Enable OpenTelemetry tracing"
            echo "  -h, --help           Show this help message"
            echo ""
35
            echo "Any additional options are passed through to dynamo.trtllm."
36
37
38
            exit 0
            ;;
        *)
39
40
            EXTRA_ARGS+=("$1")
            shift
41
42
43
44
45
46
47
48
49
50
51
52
            ;;
    esac
done

TRACE_ARGS=()
if [ "$ENABLE_OTEL" = true ]; then
    export DYN_LOGGING_JSONL=true
    export OTEL_EXPORT_ENABLED=1
    export OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=${OTEL_EXPORTER_OTLP_TRACES_ENDPOINT:-http://localhost:4317}
    TRACE_ARGS+=(--override-engine-args "{\"return_perf_metrics\": true, \"otlp_traces_endpoint\": \"${OTEL_EXPORTER_OTLP_TRACES_ENDPOINT}\" }")
fi

53

54
# run frontend
55
# dynamo.frontend accepts either --http-port flag or DYN_HTTP_PORT env var (defaults to 8000)
56
OTEL_SERVICE_NAME=dynamo-frontend \
57
python3 -m dynamo.frontend &
58
59

# run worker
60
# Additional command line args can be passed
61
OTEL_SERVICE_NAME=dynamo-worker \
62
python3 -m dynamo.trtllm \
63
64
  --model-path "$MODEL_PATH" \
  --served-model-name "$SERVED_MODEL_NAME" \
65
  --modality "$MODALITY" \
66
  --extra-engine-args "$AGG_ENGINE_ARGS" \
67
  "${TRACE_ARGS[@]}" \
68
69
70
71
  "${EXTRA_ARGS[@]}" &

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