debug(lipsync): v3 - docker cp python scripts, check nginx/DB

.gitea/workflows/debug-lipsync-diag.yml

@@ -0,0 +1,43 @@
+name: Debug Lipsync SSH Diag
+on:
+  workflow_dispatch:
+    inputs:
+      reason:
+        description: "debug reason"
+        required: false
+        default: "lipsync tts_processing stuck diagnosis"
+  push:
+    branches:
+      - debug/lipsync-ssh-diag
+permissions:
+  contents: read
+jobs:
+  ssh-diag:
+    name: SSH Staging Diagnostics
+    runs-on: runtime-builder
+    timeout-minutes: 15
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Setup SSH
+        shell: bash
+        env:
+          STAGING_SSH_KEY: ${{ secrets.PREVIEW_SSH_KEY }}
+        run: |
+          set -euo pipefail
+          which ssh || (apt-get update -qq && apt-get install -y -qq openssh-client)
+          mkdir -p ~/.ssh && chmod 700 ~/.ssh
+          printf "%s" "$STAGING_SSH_KEY" > ~/.ssh/id_rsa
+          chmod 600 ~/.ssh/id_rsa
+          H=47.98.113.167; P=22222
+          ssh-keyscan -p $P -H $H >> ~/.ssh/known_hosts 2>/dev/null
+          ssh -p $P -i ~/.ssh/id_rsa -o StrictHostKeyChecking=no root@$H "echo SSH_OK; hostname; date"
+      - name: Upload diag script and run
+        shell: bash
+        run: |
+          set -euo pipefail
+          H=47.98.113.167; P=22222
+          scp -P $P -i ~/.ssh/id_rsa -o StrictHostKeyChecking=no \
+            infra/scripts/server-lipsync-diag-v3.sh root@$H:/tmp/server-lipsync-diag-v3.sh
+          ssh -p $P -i ~/.ssh/id_rsa -o StrictHostKeyChecking=no root@$H \
+            "bash /tmp/server-lipsync-diag-v3.sh 2>&1"

infra/scripts/server-lipsync-diag-v2.sh

@@ -0,0 +1,152 @@
+#!/bin/bash
+# Lipsync 深度诊断脚本 v2 - 重点排查broker连通/消息投递/DB状态
+set +e
+echo "#########################################################"
+echo "#  Lipsync 深度诊断 v2"
+echo "#  Date: $(date)"
+echo "#########################################################"
+
+W=$(docker ps --format '{{.Names}}' | grep -E 'worker' | head -1)
+A=$(docker ps --format '{{.Names}}' | grep -E 'api' | grep -v 'web' | head -1)
+R=$(docker ps --format '{{.Names}}' | grep -E 'redis' | head -1)
+echo "Containers: W=$W  A=$A  R=$R"
+
+echo ""
+echo "=== 1. 容器 uptime + image tag ==="
+for c in $W $A; do
+  echo "--- $c ---"
+  docker inspect "$c" --format 'Image={{.Config.Image}} Created={{.Created}} Started={{.State.StartedAt}} Restarts={{.RestartCount}}'
+done
+
+echo ""
+echo "=== 2. API+Worker 的 BROKER/REDIS 地址（隐藏密码） ==="
+echo "--- API BROKER ---"
+docker exec "$A" env | grep -iE "broker|celery_broker|redis_url|backend" | sed -E 's|(:)?//[^:]+:([^@]+)@|\1//***:***@|g'
+echo "--- Worker BROKER ---"
+docker exec "$W" env | grep -iE "broker|celery_broker|redis_url|backend" | sed -E 's|(:)?//[^:]+:([^@]+)@|\1//***:***@|g'
+
+echo ""
+echo "=== 3. API容器 完整日志（近1000行） lipsync/TTS 相关 ==="
+docker logs --tail=1000 "$A" 2>&1 | grep -iE "lipsync|tts_synth|tts_processing|celery.*task|apply_async|NotRegistered|OperationalError|ConnectionError|error.*submit|traceback|submitted" | tail -200
+
+echo ""
+echo "=== 4. API 容器最近的 ERROR/Exception ==="
+docker logs --tail=2000 "$A" 2>&1 | grep -iE "error|exception|traceback|critical" | grep -v "health\|/health" | tail -80
+
+echo ""
+echo "=== 5. 测试1：在API容器内实际投递一条测试消息，看Worker是否消费 ==="
+# 投递后立即检查队列和Worker日志
+docker exec "$A" python - <<'PYEOF'
+import sys, time, traceback
+try:
+    from app.tasks.lipsync_tts import tts_synthesize_and_submit
+    from app.core.celery_app import celery_app as api_app
+    # 检查broker连接
+    conn = api_app.connection()
+    conn.ensure_connection(max_retries=2)
+    print("API celery broker connected:", conn.as_uri())
+    # 投递到celery默认队列
+    result = tts_synthesize_and_submit.apply_async(
+        args=["diag-test-job-id", "diag-user-id", "diag-voice", "diagnostic script text", 1.0, "neutral"],
+        queue="celery",
+    )
+    print("APPLY_ASYNC_OK task_id:", result.id)
+    print("task name:", result.name)
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+echo ""
+sleep 3
+echo "--- After apply_async: queue lengths ---"
+for q in celery generation transcode; do
+    echo "  $q: $(docker exec "$R" redis-cli LLEN $q)"
+done
+
+echo ""
+echo "=== 6. 等8秒看Worker是否收到并消费 ==="
+sleep 8
+echo "--- Queue lengths after 8s ---"
+for q in celery generation transcode; do
+    echo "  $q: $(docker exec "$R" redis-cli LLEN $q)"
+done
+
+echo ""
+echo "=== 7. Worker 日志最新记录（查看是否消费了测试消息） ==="
+docker logs --since=30s "$W" 2>&1 | tail -60
+
+echo ""
+echo "=== 8. Worker 内 Python 直接连接broker测试 ==="
+docker exec "$W" python - <<'PYEOF'
+import traceback
+try:
+    from worker_app.celery_app import celery_app
+    conn = celery_app.connection()
+    conn.ensure_connection(max_retries=2)
+    print("Worker celery broker connected:", conn.as_uri())
+    insp = celery_app.control.inspect(timeout=3)
+    reg = insp.registered()
+    print("Registered from worker inspect (via broker):")
+    for node, tasks in (reg or {}).items():
+        has_l = any('lipsync' in t for t in tasks)
+        print(f"  {node}: {len(tasks)} tasks, lipsync registered: {has_l}")
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+
+echo ""
+echo "=== 9. DB 查询最近10条lipsync_jobs状态 ==="
+docker exec "$A" python - <<'PYEOF' 2>&1
+import os, traceback
+db_url = os.environ.get("DATABASE_URL","")
+print("DATABASE_URL prefix:", (db_url[:50]+"...") if db_url else "(empty)")
+try:
+    # 通过现有代码路径
+    from app.db.session import SessionLocal
+    from packages.adapters.sqlalchemy_impl.models import LipsyncJobModel
+    db = SessionLocal()
+    jobs = db.query(LipsyncJobModel).order_by(LipsyncJobModel.created_at.desc()).limit(10).all()
+    print(f"Found {len(jobs)} recent lipsync jobs:")
+    for j in jobs:
+        err = getattr(j, 'error_message', '') or ''
+        t_id = getattr(j, 'celery_task_id', '') or ''
+        print(f"  id={j.id} status={j.status} mode={getattr(j,'mode','?')} "
+              f"created={j.created_at} celery_task_id={t_id} "
+              f"error={(err[:120]+'...') if len(err)>120 else err!r}")
+    db.close()
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+
+echo ""
+echo "=== 10. Worker 启动完整日志（前60行）==="
+docker logs "$W" 2>&1 | head -60
+
+echo ""
+echo "=== 11. lipsync_service.py 完整 create_job 相关段 ==="
+docker exec "$A" sed -n '170,240p' /app/apps/api/app/services/lipsync_service.py 2>&1
+
+echo ""
+echo "=== 12. celery_app task_routes / queue 配置（API侧）==="
+docker exec "$A" python - <<'PYEOF'
+from app.core.celery_app import celery_app
+print("task_routes:", getattr(celery_app.conf, 'task_routes', None))
+print("task_default_queue:", celery_app.conf.task_default_queue)
+print("task_queues:", celery_app.conf.task_queues)
+print("broker_url prefix:", celery_app.conf.broker_url[:60])
+print("result_backend prefix:", str(celery_app.conf.result_backend)[:60] if celery_app.conf.result_backend else None)
+PYEOF
+
+echo ""
+echo "=== 13. Worker celery_app queue 配置 ==="
+docker exec "$W" python - <<'PYEOF'
+from worker_app.celery_app import celery_app
+print("task_routes:", getattr(celery_app.conf, 'task_routes', None))
+print("task_default_queue:", celery_app.conf.task_default_queue)
+print("task_queues:", celery_app.conf.task_queues)
+print("include/imports count:", len(celery_app.conf.imports))
+PYEOF
+
+echo ""
+echo "#########################################################"
+echo "#  v2 诊断完成"
+echo "#########################################################"

infra/scripts/server-lipsync-diag-v3.sh

@@ -0,0 +1,162 @@
+#!/bin/bash
+# Lipsync 诊断脚本 v3 - 用独立文件避免heredoc问题
+set +e
+echo "#########################################################"
+echo "#  Lipsync 诊断 v3"
+echo "#  Date: $(date)"
+echo "#########################################################"
+
+W=$(docker ps --format '{{.Names}}' | grep -E 'worker' | head -1)
+A=$(docker ps --format '{{.Names}}' | grep -E 'api' | grep -v 'web' | head -1)
+R=$(docker ps --format '{{.Names}}' | grep -E 'redis' | head -1)
+NGINX=$(docker ps --format '{{.Names}}' | grep -iE 'nginx|web' | head -1)
+echo "Containers:"
+echo "  W=$W"
+echo "  A=$A"
+echo "  R=$R"
+echo "  NGINX=$NGINX"
+
+echo ""
+echo "=== A1. API 容器近 2000 行日志全部（看请求是否进来） ==="
+docker logs --tail=2000 "$A" 2>&1 | tail -300
+
+echo ""
+echo "=== A2. API容器启动命令 & 网络 ==="
+docker inspect "$A" --format 'Cmd={{.Config.Cmd}} Entrypoint={{.Config.Entrypoint}} NetworkMode={{.HostConfig.NetworkMode}}'
+docker inspect "$A" --format '{{range .NetworkSettings.Networks}}{{.NetworkID}} {{.IPAddress}}{{end}}'
+
+echo ""
+echo "=== A3. Nginx 容器（如果有）日志 ==="
+if [ -n "$NGINX" ]; then
+  docker logs --tail=200 "$NGINX" 2>&1 | grep -iE "lipsync|ai-avatar|error|upstream" | tail -100
+  echo "--- nginx config ---"
+  docker exec "$NGINX" cat /etc/nginx/conf.d/default.conf 2>&1 | head -80
+fi
+
+echo ""
+echo "=== B1. 写测试脚本到/tmp，再docker cp到容器里执行（避免heredoc问题） ==="
+cat > /tmp/api_diag.py << 'PYEOF'
+import sys, traceback, os
+print("=== PYTHON OK in API container, cwd=", os.getcwd())
+print("sys.path[0:3]:", sys.path[:3])
+print("PYTHONPATH:", os.environ.get("PYTHONPATH",""))
+try:
+    from app.tasks.lipsync_tts import tts_synthesize_and_submit
+    from app.core.celery_app import celery_app as api_app
+    print("api_app.main:", api_app.main)
+    print("api_app.conf.broker_url:", api_app.conf.broker_url)
+    print("task name:", tts_synthesize_and_submit.name)
+    print("task.app.main:", tts_synthesize_and_submit.app.main)
+    print("task in api_app.tasks:", tts_synthesize_and_submit.name in api_app.tasks)
+    print("api_app.conf.task_routes:", api_app.conf.task_routes)
+    print("api_app.conf.task_default_queue:", api_app.conf.task_default_queue)
+    # 测试broker连接
+    conn = api_app.connection()
+    conn.ensure_connection(max_retries=2)
+    print("broker connected OK:", conn.as_uri())
+    # 投递测试任务
+    result = tts_synthesize_and_submit.apply_async(
+        args=["diag-v3-job", "diag-user", "diag-voice", "hello v3 diag", 1.0, "neutral"],
+        queue="celery",
+    )
+    print("APPLY_ASYNC_OK id=", result.id, "name=", result.name, "queue=celery")
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+docker cp /tmp/api_diag.py "$A:/tmp/api_diag.py"
+echo "--- docker exec python /tmp/api_diag.py in API ---"
+docker exec "$A" python /tmp/api_diag.py 2>&1
+
+echo ""
+echo "=== B2. 投递3秒后查队列 ==="
+sleep 3
+for q in celery generation transcode; do
+    echo "  queue $q: $(docker exec "$R" redis-cli LLEN $q)"
+done
+echo "--- peek celery ---"
+docker exec "$R" redis-cli LRANGE celery 0 2 2>&1 | head -20
+
+echo ""
+echo "=== B3. 等10秒再查队列和Worker日志 ==="
+sleep 7
+for q in celery generation transcode; do
+    echo "  queue $q: $(docker exec "$R" redis-cli LLEN $q)"
+done
+echo "--- Worker logs since 20s ago ---"
+docker logs --since=20s "$W" 2>&1 | tail -50
+
+echo ""
+echo "=== C1. Worker侧执行诊断脚本 ==="
+cat > /tmp/worker_diag.py << 'PYEOF'
+import sys, os, traceback
+print("=== PYTHON OK in Worker container ===")
+try:
+    from worker_app.celery_app import celery_app
+    print("worker celery_app.main:", celery_app.main)
+    print("broker_url:", celery_app.conf.broker_url)
+    print("task_default_queue:", celery_app.conf.task_default_queue)
+    print("task_routes:", celery_app.conf.task_routes)
+    print("task_queues:", [(q.name, [b.name for b in q.bindings]) for q in (celery_app.conf.task_queues or [])])
+    # inspect via broker
+    insp = celery_app.control.inspect(timeout=3)
+    act = insp.active() or {}
+    reg = insp.registered() or {}
+    res = insp.reserved() or {}
+    for node, tasks in act.items():
+        print(f"ACTIVE on {node}: {len(tasks)} tasks")
+        for t in tasks:
+            print(f"  - {t.get('name')} args={str(t.get('args',''))[:60]}")
+    for node, tasks in res.items():
+        print(f"RESERVED on {node}: {len(tasks)} tasks")
+        for t in tasks:
+            print(f"  - {t.get('name')}")
+    print("registered lipsync count per node:")
+    for node, tasks in reg.items():
+        has = [t for t in tasks if 'lipsync' in t]
+        print(f"  {node}: total={len(tasks)}, lipsync={has}")
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+docker cp /tmp/worker_diag.py "$W:/tmp/worker_diag.py"
+docker exec "$W" python /tmp/worker_diag.py 2>&1
+
+echo ""
+echo "=== D1. DB最近20条lipsync jobs ==="
+cat > /tmp/db_diag.py << 'PYEOF'
+import sys, os, traceback
+print("=== DB diag in API container ===")
+print("DATABASE_URL prefix:", os.environ.get("DATABASE_URL","")[:60])
+try:
+    from app.db.session import SessionLocal
+    from packages.adapters.sqlalchemy_impl.models import LipsyncJobModel
+    from datetime import datetime, timezone
+    db = SessionLocal()
+    jobs = db.query(LipsyncJobModel).order_by(LipsyncJobModel.created_at.desc()).limit(20).all()
+    print(f"Found {len(jobs)} recent lipsync jobs:")
+    for j in jobs:
+        err = getattr(j, 'error_message', None) or ''
+        t_id = getattr(j, 'celery_task_id', None) or ''
+        c_at = j.created_at.isoformat() if j.created_at else '?'
+        print(f"  id={j.id} status={j.status} mode={'tts' if (j.voice_id and not j.audio_url) else 'audio'} "
+              f"created={c_at} celery_task_id={t_id} err={(err[:100]+'...') if len(err)>100 else err!r}")
+    db.close()
+except Exception as e:
+    traceback.print_exc()
+PYEOF
+docker cp /tmp/db_diag.py "$A:/tmp/db_diag.py"
+docker exec "$A" python /tmp/db_diag.py 2>&1
+
+echo ""
+echo "=== D2. API容器监听端口 ==="
+docker exec "$A" sh -c "ss -tlnp 2>/dev/null || netstat -tlnp 2>/dev/null || cat /proc/net/tcp | head" 2>&1 | head -20
+
+echo ""
+echo "=== E1. 外部HTTPS健康检查 ==="
+curl -sk --max-time 8 https://staging-api.xiaoxiajianji.com/api/v1/health 2>&1 | head -3
+echo ""
+curl -sk --max-time 8 https://staging-api.xiaoxiajianji.com/api/v1/lipsync/jobs 2>&1 | head -3
+
+echo ""
+echo "#########################################################"
+echo "#  v3 诊断完成"
+echo "#########################################################"

infra/scripts/server-lipsync-diag.sh

@@ -0,0 +1,117 @@
+#!/bin/bash
+# Lipsync staging 诊断脚本（在staging服务器本机执行）
+set +e
+echo "#########################################################"
+echo "#  Lipsync TTS async 诊断报告"
+echo "#  Host: $(hostname)"
+echo "#  Date: $(date)"
+echo "#########################################################"
+
+echo ""
+echo "=== 1. docker ps (worker/api/redis) ==="
+docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}\t{{.CreatedAt}}' | grep -E "worker|api|redis|nginx|NAMES"
+
+W=$(docker ps --format '{{.Names}}' | grep -E 'worker' | head -1)
+A=$(docker ps --format '{{.Names}}' | grep -E 'api' | grep -v 'web' | head -1)
+R=$(docker ps --format '{{.Names}}' | grep -E 'redis' | head -1)
+
+echo ""
+echo "=== Worker container: $W ==="
+echo "=== API container: $A ==="
+echo "=== Redis container: $R ==="
+
+echo ""
+echo "=== 2. Worker container image + created time ==="
+docker inspect "$W" --format 'Image={{.Config.Image}} Created={{.Created}}'
+docker inspect "$A" --format 'Image={{.Config.Image}} Created={{.Created}}'
+
+echo ""
+echo "=== 3. Worker 启动日志：celery ready / registered / lipsync 相关 ==="
+docker logs --tail=500 "$W" 2>&1 | grep -iE "lipsync|tts_synth|synthesize|celery@|ready|registered|register|error|traceback|import|critical|not registered|Consumer|mingle|tasks" | tail -120
+
+echo ""
+echo "=== 4. celery -A inspect registered ==="
+docker exec "$W" celery -A worker_app.celery_app inspect registered 2>&1 | tail -80
+
+echo ""
+echo "=== 5. celery -A inspect active/reserved ==="
+echo "--- active ---"
+docker exec "$W" celery -A worker_app.celery_app inspect active 2>&1 | tail -30
+echo "--- reserved ---"
+docker exec "$W" celery -A worker_app.celery_app inspect reserved 2>&1 | tail -30
+
+echo ""
+echo "=== 6. Worker 内 Python: import lipsync_tts task，看绑定 ==="
+docker exec "$W" python -c "
+from worker_app.celery_app import celery_app
+print('worker celery_app.main:', celery_app.main)
+print('worker celery_app.conf.imports:')
+for m in celery_app.conf.imports:
+    print('  ', m)
+import app.tasks.lipsync_tts as m
+t = m.tts_synthesize_and_submit
+print('task name:', t.name)
+print('task.app.main:', t.app.main)
+print('task.app is worker celery_app:', t.app is celery_app)
+print('task.name in worker celery_app.tasks:', t.name in celery_app.tasks)
+" 2>&1
+
+echo ""
+echo "=== 7. API 容器日志：lipsync/celery/apply_async 相关错误 ==="
+docker logs --tail=800 "$A" 2>&1 | grep -iE "lipsync|tts_|celery|apply_async|NotRegistered|traceback|error.*task" | tail -120
+
+echo ""
+echo "=== 8. API 内 Python: import task 看绑定 + apply_async 试投递（不消费） ==="
+docker exec "$A" python -c "
+import traceback
+try:
+    from app.tasks.lipsync_tts import tts_synthesize_and_submit
+    from app.core.celery_app import celery_app as api_app
+    print('api celery_app.main:', api_app.main)
+    print('task name:', tts_synthesize_and_submit.name)
+    print('task.app.main:', tts_synthesize_and_submit.app.main)
+    print('task.app is api_app:', tts_synthesize_and_submit.app is api_app)
+    print('task.name in api_app.tasks:', tts_synthesize_and_submit.name in api_app.tasks)
+    # 测试 send_task 是否能路由（不发送真任务）
+    print('api_app tasks includes lipsync_tts.synthesize_and_submit:', 'lipsync_tts.synthesize_and_submit' in api_app.tasks)
+except Exception as e:
+    traceback.print_exc()
+" 2>&1
+
+echo ""
+echo "=== 9. Redis 队列长度 ==="
+for q in celery generation transcode; do
+    len=$(docker exec "$R" redis-cli LLEN $q 2>&1)
+    echo "  $q length: $len"
+done
+
+echo ""
+echo "=== 10. Redis celery 队列 peek（最多5条） ==="
+docker exec "$R" redis-cli LRANGE celery 0 4 2>&1 | head -40
+
+echo ""
+echo "=== 11. Worker 关键文件内容校验 ==="
+echo "--- worker_app/celery_app.py imports 段 ---"
+docker exec "$W" grep -n "lipsync\|imports\s*=\|app.tasks\|apps.api" /app/apps/worker/worker_app/celery_app.py 2>&1
+echo ""
+echo "--- lipsync_tts.py 前20行（应该是 shared_task） ---"
+docker exec "$W" head -25 /app/apps/api/app/tasks/lipsync_tts.py 2>&1
+
+echo ""
+echo "=== 12. API lipsync_service.py apply_async 上下文 ==="
+docker exec "$A" grep -n -B3 -A10 "apply_async\|tts_synthesize" /app/apps/api/app/services/lipsync_service.py 2>&1 | head -80
+
+echo ""
+echo "=== 13. Worker /app 目录结构（app/tasks） ==="
+docker exec "$W" ls -la /app/apps/api/app/tasks/ 2>&1 | head -30
+echo ""
+echo "--- /app/apps/api/app/__init__.py? ---"
+docker exec "$W" ls -la /app/apps/api/app/__init__.py /app/apps/api/app/core/__init__.py /app/apps/api/app/api/__init__.py 2>&1
+echo ""
+echo "--- PYTHONPATH inside worker ---"
+docker exec "$W" env | grep PYTHONPATH
+
+echo ""
+echo "#########################################################"
+echo "#  诊断完成"
+echo "#########################################################"