start script: llama-server + health wait + uvicorn

start.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+set -euo pipefail
+
+MODEL=/app/models/semanticwiki-coder-7b-v2-q4_k_m.gguf
+CTX="${CONTEXT_SIZE:-16384}"
+
+echo "[start] launching llama-server on :8080 (ctx=$CTX)"
+llama-server \
+    -m "$MODEL" \
+    --host 0.0.0.0 \
+    --port 8080 \
+    -c "$CTX" \
+    -ngl 999 \
+    --jinja &
+LLAMA_PID=$!
+
+echo "[start] waiting for llama-server health..."
+for i in $(seq 1 180); do
+    if curl -sf http://127.0.0.1:8080/health > /dev/null; then
+        echo "[start] llama-server healthy after ${i}s"
+        break
+    fi
+    if ! kill -0 "$LLAMA_PID" 2>/dev/null; then
+        echo "[start] llama-server exited early"
+        exit 1
+    fi
+    sleep 1
+done
+
+exec python -m uvicorn main:app --host 0.0.0.0 --port 7860