ProjectParrot/setup.sh at main · LofiChptr6/ProjectParrot · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
#!/usr/bin/env bash
set -euo pipefail

# ---------------------------------------------------------------------------
# ProjectParrot — First-time setup
#
# GPU-heavy services (vLLM) run in Docker with explicit GPU reservation.
# Python services (STT, TTS, memory, animation, bridge) run natively.
# ---------------------------------------------------------------------------

PRO6000_GPU_INDEX=0

green()  { printf '\033[0;32m%s\033[0m\n' "$*"; }
yellow() { printf '\033[0;33m%s\033[0m\n' "$*"; }
red()    { printf '\033[0;31m%s\033[0m\n' "$*"; }

cd "$(dirname "$0")"

# ── 1. System dependencies ─────────────────────────────────────────
green "==> Installing system dependencies…"
sudo dnf install -y zstd curl 2>/dev/null \
  || sudo apt-get update -qq && sudo apt-get install -y zstd curl 2>/dev/null \
  || yellow "  Could not install zstd/curl — install them manually if missing."

# ── 2. Verify GPU visibility ───────────────────────────────────────
green "==> Checking NVIDIA GPUs…"
if ! command -v nvidia-smi &>/dev/null; then
    red "  nvidia-smi not found."
    red "  Install the NVIDIA driver for your distro (e.g. akmod-nvidia on Fedora,"
    red "  nvidia-utils-570 on Ubuntu/Debian)."
    exit 1
fi
nvidia-smi -L
echo

# ── 3. Verify Docker + NVIDIA Container Toolkit ───────────────────
green "==> Checking Docker…"
if ! command -v docker &>/dev/null; then
    red "  Docker not found. Install Docker Engine first:"
    red "    https://docs.docker.com/engine/install/"
    exit 1
fi
green "  Docker: $(docker --version)"

green "==> Checking NVIDIA Container Toolkit…"
if ! docker info 2>/dev/null | grep -qi "nvidia"; then
    yellow "  NVIDIA runtime not detected in Docker."
    yellow "  Quick test: running a GPU container…"
    if docker run --rm --gpus device=0 nvidia/cuda:12.8.1-base-ubuntu24.04 nvidia-smi &>/dev/null; then
        green "  GPU access works despite missing 'nvidia' in docker info — OK."
    else
        red ""
        red "  Docker cannot access GPUs. Install NVIDIA Container Toolkit:"
        red ""
        red "  Fedora/RHEL:"
        red "    curl -s -L https://nvidia.github.io/libnvidia-container/stable/rpm/nvidia-container-toolkit.repo \\"
        red "      | sudo tee /etc/yum.repos.d/nvidia-container-toolkit.repo"
        red "    sudo dnf install -y nvidia-container-toolkit"
        red "    sudo nvidia-ctk runtime configure --runtime=docker"
        red "    sudo systemctl restart docker"
        red ""
        red "  Ubuntu/Debian:"
        red "    curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey \\"
        red "      | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg"
        red "    curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list \\"
        red "      | sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' \\"
        red "      | sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list"
        red "    sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit"
        red "    sudo nvidia-ctk runtime configure --runtime=docker"
        red "    sudo systemctl restart docker"
        red ""
        red "  Then re-run this script."
        exit 1
    fi
else
    green "  NVIDIA Container Toolkit: detected."
fi
echo

# ── 4. Disable native Ollama (if installed) ────────────────────────
if systemctl is-active --quiet ollama 2>/dev/null; then
    yellow "==> Stopping native Ollama systemd service (no longer needed)…"
    sudo systemctl stop ollama
    sudo systemctl disable ollama
    green "  Native Ollama disabled."
fi
echo

# ── 5. Start vLLM in Docker ──────────────────────────────────────
green "==> Starting vLLM in Docker (GPU ${PRO6000_GPU_INDEX})…"
docker compose pull
docker compose up -d

# Wait for health check (model download + load can take several minutes)
yellow "  Waiting for vLLM to become healthy (this may take a few minutes on first run)…"
for i in $(seq 1 120); do
    if curl -sf --connect-timeout 2 http://127.0.0.1:8800/health >/dev/null 2>&1; then
        green "  vLLM is up."
        break
    fi
    if [[ "$i" -eq 120 ]]; then
        red "  vLLM did not respond after 120s. Check: docker compose logs vllm"
        exit 1
    fi
    sleep 2
done
echo

# ── 6. Verify GPU is actually being used ──────────────────────────
green "==> Verifying GPU access inside container…"
if docker exec vllm nvidia-smi &>/dev/null; then
    green "  Container has GPU access:"
    docker exec vllm nvidia-smi --query-gpu=index,name,memory.total --format=csv,noheader
else
    red "  WARNING: Container does NOT have GPU access!"
    red "  vLLM will fail to load models. Fix NVIDIA Container Toolkit and re-run."
fi
echo

# ── 7. Verify model is loaded ────────────────────────────────────
green "==> Checking loaded models…"
curl -s http://127.0.0.1:8800/v1/models | python3 -c "
import sys, json
data = json.load(sys.stdin)
for m in data.get('data', []):
    print(f'  Model: {m[\"id\"]}')
" 2>/dev/null || yellow "  Could not list models — vLLM may still be loading."
echo

# ── 8. Prepare .env ──────────────────────────────────────────────
if [ ! -f .env ]; then
    cp .env.example .env
    yellow "  Created .env from .env.example — set HF_TOKEN for gated models."
else
    yellow "  .env already exists, skipping."
fi
echo

# ── Done ──────────────────────────────────────────────────────────
green "╔══════════════════════════════════════════════════════════════╗"
green "║  ProjectParrot setup complete!                              ║"
green "║                                                             ║"
green "║  vLLM (Docker):  http://127.0.0.1:8800  (GPU ${PRO6000_GPU_INDEX})         ║"
green "║                                                             ║"
green "║  Next: ./start.sh all                                       ║"
green "╚══════════════════════════════════════════════════════════════╝"