更新 start.sh

start.sh

@@ -1,11 +1,11 @@
 #!/usr/bin/env bash
 set -euo pipefail
 
-# ── 1. Validate required LiteLLM variables ────────────────────────────────────
-: "${LITELLM_API_KEY:?LITELLM_API_KEY secret is required}"
-: "${LITELLM_MODEL:?LITELLM_MODEL secret is required (e.g. openai/gpt-4o)}"
+# ── 1. Validate required variables ────────────────────────────────────────────
+: "${LITELLM_API_KEY:?Secret LITELLM_API_KEY is required}"
+: "${LITELLM_MODEL:?Secret LITELLM_MODEL is required (e.g. openai/gpt-4o or anthropic/claude-3-5-sonnet-20241022)}"
 
-# ── 2. Generate LiteLLM proxy config dynamically ──────────────────────────────
+# ── 2. Write LiteLLM proxy config ─────────────────────────────────────────────
 LITELLM_CONFIG=/tmp/litellm_config.yaml
 
 {
@@ -24,23 +24,23 @@ LITELLM_CONFIG=/tmp/litellm_config.yaml
   echo "  request_timeout: 120"
 } > "$LITELLM_CONFIG"
 
-echo "[start.sh] LiteLLM config written:"
+echo "[start.sh] LiteLLM config:"
 cat "$LITELLM_CONFIG"
 
 # ── 3. Start LiteLLM proxy in the background ──────────────────────────────────
 litellm --config "$LITELLM_CONFIG" --port 4000 --host 127.0.0.1 &
 LITELLM_PID=$!
-echo "[start.sh] LiteLLM proxy started (pid $LITELLM_PID)"
+echo "[start.sh] LiteLLM proxy started (pid=$LITELLM_PID)"
 
-# Forward SIGTERM to LiteLLM when we exit
-trap 'kill $LITELLM_PID 2>/dev/null || true' EXIT TERM INT
+# Kill LiteLLM on any exit
+trap 'echo "[start.sh] shutting down LiteLLM"; kill "$LITELLM_PID" 2>/dev/null || true' EXIT TERM INT
 
-# ── 4. Wait for LiteLLM to become healthy (max 30 s) ─────────────────────────
-MAX_WAIT=30
+# ── 4. Wait for LiteLLM to be healthy (max 60 s) ─────────────────────────────
+MAX_WAIT=60
 WAITED=0
 until curl -sf http://127.0.0.1:4000/health/liveliness > /dev/null 2>&1; do
     if [ "$WAITED" -ge "$MAX_WAIT" ]; then
-        echo "[start.sh] ERROR: LiteLLM did not become healthy within ${MAX_WAIT}s — aborting"
+        echo "[start.sh] ERROR: LiteLLM not healthy after ${MAX_WAIT}s"
         exit 1
     fi
     sleep 1
@@ -48,9 +48,10 @@ until curl -sf http://127.0.0.1:4000/health/liveliness > /dev/null 2>&1; do
 done
 echo "[start.sh] LiteLLM healthy after ${WAITED}s"
 
-# ── 5. Launch OpenClaw gateway pointing at the LiteLLM proxy ─────────────────
-# We expose LiteLLM as an OpenAI-compatible endpoint on 127.0.0.1:4000.
-# OPENCLAW_DEFAULT_MODEL must match the model_name defined above ("default").
+# ── 5. Start OpenClaw gateway pointing at LiteLLM ─────────────────────────────
+# OPENAI_API_KEY / OPENAI_BASE_URL: tell OpenClaw to use LiteLLM as its backend.
+# OPENCLAW_DEFAULT_MODEL=default:   matches the model_name in litellm_config.yaml.
+# OPENCLAW_API_PORT is already set to 7860 in the Dockerfile ENV.
 exec env \
     OPENCLAW_DEFAULT_MODEL=default \
     OPENAI_API_KEY=litellm-proxy \