[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C0QH3WG8SH3DK6AS9EEDCQ":3,"skill-same:01M7C0QH3WG8SH3DK6AS9EEDCQ":50},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":32,"ownerHtmlUrl":33,"ownerBlogUrl":6,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M7C0QH3WG8SH3DK6AS9EEDCQ","gcp-cloud-run",null,"Specialized skill for building production-ready serverless applications on GCP. Covers Cloud Run services (containerized), Cloud Run Functions (event-driven), cold start optimization, and event-driven architecture with Pub\u002FSub.","YYC-Cube\u002FYYC3-Knowledge-Base","YYC-Cube","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F235006543?v=4","01M79J2PH4SWD2X0825X6PN748","yyc3","01M79J2PHVK3F6ZBYT8K2SCZCW",1,"MIT","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-antigravity-awesome-skills\u002Fplugins\u002Fantigravity-awesome-skills-claude\u002Fskills\u002Fgcp-cloud-run\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: gcp-cloud-run\ndescription: Specialized skill for building production-ready serverless\n  applications on GCP. Covers Cloud Run services (containerized), Cloud Run\n  Functions (event-driven), cold start optimization, and event-driven\n  architecture with Pub\u002FSub.\nrisk: unknown\nsource: vibeship-spawner-skills (Apache 2.0)\ndate_added: 2026-02-27\n---\n\n# GCP Cloud Run\n\nSpecialized skill for building production-ready serverless applications on GCP.\nCovers Cloud Run services (containerized), Cloud Run Functions (event-driven),\ncold start optimization, and event-driven architecture with Pub\u002FSub.\n\n## Principles\n\n- Cloud Run for containers, Functions for simple event handlers\n- Optimize for cold starts with startup CPU boost and min instances\n- Set concurrency based on workload (start with 8, adjust)\n- Memory includes \u002Ftmp filesystem - plan accordingly\n- Use VPC Connector only when needed (adds latency)\n- Containers should start fast and be stateless\n- Handle signals gracefully for clean shutdown\n\n## Patterns\n\n### Cloud Run Service Pattern\n\nContainerized web service on Cloud Run\n\n**When to use**: Web applications and APIs,Need any runtime or library,Complex services with multiple endpoints,Stateless containerized workloads\n\n```dockerfile\n# Dockerfile - Multi-stage build for smaller image\nFROM node:20-slim AS builder\nWORKDIR \u002Fapp\nCOPY package*.json .\u002F\nRUN npm ci --only=production\n\nFROM node:20-slim\nWORKDIR \u002Fapp\n\n# Copy only production dependencies\nCOPY --from=builder \u002Fapp\u002Fnode_modules .\u002Fnode_modules\nCOPY src .\u002Fsrc\nCOPY package.json .\u002F\n\n# Cloud Run uses PORT env variable\nENV PORT=8080\nEXPOSE 8080\n\n# Run as non-root user\nUSER node\n\nCMD [\"node\", \"src\u002Findex.js\"]\n```\n\n```javascript\n\u002F\u002F src\u002Findex.js\nconst express = require('express');\nconst app = express();\n\napp.use(express.json());\n\n\u002F\u002F Health check endpoint\napp.get('\u002Fhealth', (req, res) => {\n  res.status(200).send('OK');\n});\n\n\u002F\u002F API routes\napp.get('\u002Fapi\u002Fitems\u002F:id', async (req, res) => {\n  try {\n    const item = await getItem(req.params.id);\n    res.json(item);\n  } catch (error) {\n    console.error('Error:', error);\n    res.status(500).json({ error: 'Internal server error' });\n  }\n});\n\n\u002F\u002F Graceful shutdown\nprocess.on('SIGTERM', () => {\n  console.log('SIGTERM received, shutting down gracefully');\n  server.close(() => {\n    console.log('Server closed');\n    process.exit(0);\n  });\n});\n\nconst PORT = process.env.PORT || 8080;\nconst server = app.listen(PORT, () => {\n  console.log(`Server listening on port ${PORT}`);\n});\n```\n\n```yaml\n# cloudbuild.yaml\nsteps:\n  # Build the container image\n  - name: 'gcr.io\u002Fcloud-builders\u002Fdocker'\n    args: ['build', '-t', 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA', '.']\n\n  # Push the container image\n  - name: 'gcr.io\u002Fcloud-builders\u002Fdocker'\n    args: ['push', 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA']\n\n  # Deploy to Cloud Run\n  - name: 'gcr.io\u002Fgoogle.com\u002Fcloudsdktool\u002Fcloud-sdk'\n    entrypoint: gcloud\n    args:\n      - 'run'\n      - 'deploy'\n      - 'my-service'\n      - '--image=gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA'\n      - '--region=us-central1'\n      - '--platform=managed'\n      - '--allow-unauthenticated'\n      - '--memory=512Mi'\n      - '--cpu=1'\n      - '--min-instances=1'\n      - '--max-instances=100'\n      - '--concurrency=80'\n      - '--cpu-boost'\n\nimages:\n  - 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA'\n```\n\n### Structure\n\nproject\u002F\n├── Dockerfile\n├── .dockerignore\n├── src\u002F\n│   ├── index.js\n│   └── routes\u002F\n├── package.json\n└── cloudbuild.yaml\n\n### Gcloud_deploy\n\n# Direct gcloud deployment\ngcloud run deploy my-service \\\n  --source . \\\n  --region us-central1 \\\n  --allow-unauthenticated \\\n  --memory 512Mi \\\n  --cpu 1 \\\n  --min-instances 1 \\\n  --max-instances 100 \\\n  --concurrency 80 \\\n  --cpu-boost\n\n### Cloud Run Functions Pattern\n\nEvent-driven functions (formerly Cloud Functions)\n\n**When to use**: Simple event handlers,Pub\u002FSub message processing,Cloud Storage triggers,HTTP webhooks\n\n```javascript\n\u002F\u002F HTTP Function\n\u002F\u002F index.js\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.http('helloHttp', (req, res) => {\n  const name = req.query.name || req.body.name || 'World';\n  res.send(`Hello, ${name}!`);\n});\n```\n\n```javascript\n\u002F\u002F Pub\u002FSub Function\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.cloudEvent('processPubSub', (cloudEvent) => {\n  \u002F\u002F Decode Pub\u002FSub message\n  const message = cloudEvent.data.message;\n  const data = message.data\n    ? JSON.parse(Buffer.from(message.data, 'base64').toString())\n    : {};\n\n  console.log('Received message:', data);\n\n  \u002F\u002F Process message\n  processMessage(data);\n});\n```\n\n```javascript\n\u002F\u002F Cloud Storage Function\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.cloudEvent('processStorageEvent', async (cloudEvent) => {\n  const file = cloudEvent.data;\n\n  console.log(`Event: ${cloudEvent.type}`);\n  console.log(`Bucket: ${file.bucket}`);\n  console.log(`File: ${file.name}`);\n\n  if (cloudEvent.type === 'google.cloud.storage.object.v1.finalized') {\n    await processUploadedFile(file.bucket, file.name);\n  }\n});\n```\n\n```bash\n# Deploy HTTP function\ngcloud functions deploy hello-http \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-http \\\n  --allow-unauthenticated \\\n  --region us-central1\n\n# Deploy Pub\u002FSub function\ngcloud functions deploy process-messages \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-topic my-topic \\\n  --region us-central1\n\n# Deploy Cloud Storage function\ngcloud functions deploy process-uploads \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-event-filters=\"type=google.cloud.storage.object.v1.finalized\" \\\n  --trigger-event-filters=\"bucket=my-bucket\" \\\n  --region us-central1\n```\n\n### Cold Start Optimization Pattern\n\nMinimize cold start latency for Cloud Run\n\n**When to use**: Latency-sensitive applications,User-facing APIs,High-traffic services\n\n## 1. Enable Startup CPU Boost\n\n```bash\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --region us-central1\n```\n\n## 2. Set Minimum Instances\n\n```bash\ngcloud run deploy my-service \\\n  --min-instances 1 \\\n  --region us-central1\n```\n\n## 3. Optimize Container Image\n\n```dockerfile\n# Use distroless for minimal image\nFROM node:20-slim AS builder\nWORKDIR \u002Fapp\nCOPY package*.json .\u002F\nRUN npm ci --only=production\n\nFROM gcr.io\u002Fdistroless\u002Fnodejs20-debian12\nWORKDIR \u002Fapp\nCOPY --from=builder \u002Fapp\u002Fnode_modules .\u002Fnode_modules\nCOPY src .\u002Fsrc\nCMD [\"src\u002Findex.js\"]\n```\n\n## 4. Lazy Initialize Heavy Dependencies\n\n```javascript\n\u002F\u002F Lazy load heavy libraries\nlet bigQueryClient = null;\n\nfunction getBigQueryClient() {\n  if (!bigQueryClient) {\n    const { BigQuery } = require('@google-cloud\u002Fbigquery');\n    bigQueryClient = new BigQuery();\n  }\n  return bigQueryClient;\n}\n\n\u002F\u002F Only initialize when needed\napp.get('\u002Fapi\u002Fanalytics', async (req, res) => {\n  const client = getBigQueryClient();\n  const results = await client.query({...});\n  res.json(results);\n});\n```\n\n## 5. Increase Memory (More CPU)\n\n```bash\n# Higher memory = more CPU during startup\ngcloud run deploy my-service \\\n  --memory 1Gi \\\n  --cpu 2 \\\n  --region us-central1\n```\n\n### Optimization_impact\n\n- Startup_cpu_boost: 50% faster cold starts\n- Min_instances: Eliminates cold starts for traffic spikes\n- Distroless_image: Smaller attack surface, faster pull\n- Lazy_init: Defers heavy loading to first request\n\n### Concurrency Configuration Pattern\n\nProper concurrency settings for Cloud Run\n\n**When to use**: Need to optimize instance utilization,Handle traffic spikes efficiently,Reduce cold starts\n\n## Understanding Concurrency\n\n```bash\n# Default concurrency is 80\n# Adjust based on your workload\n\n# For I\u002FO-bound workloads (most web apps)\ngcloud run deploy my-service \\\n  --concurrency 80 \\\n  --cpu 1\n\n# For CPU-bound workloads\ngcloud run deploy my-service \\\n  --concurrency 1 \\\n  --cpu 1\n\n# For memory-intensive workloads\ngcloud run deploy my-service \\\n  --concurrency 10 \\\n  --memory 2Gi\n```\n\n## Node.js Concurrency\n\n```javascript\n\u002F\u002F Node.js is single-threaded but handles I\u002FO concurrently\n\u002F\u002F Use async\u002Fawait for all I\u002FO operations\n\n\u002F\u002F GOOD - async I\u002FO\napp.get('\u002Fapi\u002Fdata', async (req, res) => {\n  const [users, products] = await Promise.all([\n    fetchUsers(),\n    fetchProducts()\n  ]);\n  res.json({ users, products });\n});\n\n\u002F\u002F BAD - blocking operation\napp.get('\u002Fapi\u002Fcompute', (req, res) => {\n  const result = heavyCpuOperation(); \u002F\u002F Blocks other requests!\n  res.json(result);\n});\n```\n\n## Python Concurrency with Gunicorn\n\n```dockerfile\nFROM python:3.11-slim\nWORKDIR \u002Fapp\nCOPY requirements.txt .\nRUN pip install --no-cache-dir -r requirements.txt\nCOPY . .\n\n# 4 workers for concurrency\nCMD exec gunicorn --bind :$PORT --workers 4 --threads 2 main:app\n```\n\n```python\n# main.py\nfrom flask import Flask\napp = Flask(__name__)\n\n@app.route('\u002Fapi\u002Fdata')\ndef get_data():\n    return {'status': 'ok'}\n```\n\n### Concurrency_guidelines\n\n- Concurrency=1: Only for CPU-bound or unsafe code\n- Concurrency=8 20: Memory-intensive workloads\n- Concurrency=80: Default, good for I\u002FO-bound\n- Concurrency=250: Maximum, for very lightweight handlers\n\n### Pub\u002FSub Integration Pattern\n\nEvent-driven processing with Cloud Pub\u002FSub\n\n**When to use**: Asynchronous message processing,Decoupled microservices,Event-driven architecture\n\n## Push Subscription to Cloud Run\n\n```bash\n# Create topic\ngcloud pubsub topics create orders\n\n# Create push subscription to Cloud Run\ngcloud pubsub subscriptions create orders-push \\\n  --topic orders \\\n  --push-endpoint https:\u002F\u002Fmy-service-xxx.run.app\u002Fpubsub \\\n  --ack-deadline 600\n```\n\n```javascript\n\u002F\u002F Handle Pub\u002FSub push messages\nconst express = require('express');\nconst app = express();\napp.use(express.json());\n\napp.post('\u002Fpubsub', async (req, res) => {\n  \u002F\u002F Verify the request is from Pub\u002FSub\n  if (!req.body.message) {\n    return res.status(400).send('Invalid Pub\u002FSub message');\n  }\n\n  try {\n    \u002F\u002F Decode message data\n    const message = req.body.message;\n    const data = message.data\n      ? JSON.parse(Buffer.from(message.data, 'base64').toString())\n      : {};\n\n    console.log('Processing order:', data);\n\n    await processOrder(data);\n\n    \u002F\u002F Return 200 to acknowledge\n    res.status(200).send('OK');\n  } catch (error) {\n    console.error('Processing failed:', error);\n    \u002F\u002F Return 500 to trigger retry\n    res.status(500).send('Processing failed');\n  }\n});\n```\n\n## Publishing Messages\n\n```javascript\nconst { PubSub } = require('@google-cloud\u002Fpubsub');\nconst pubsub = new PubSub();\n\nasync function publishOrder(order) {\n  const topic = pubsub.topic('orders');\n  const messageBuffer = Buffer.from(JSON.stringify(order));\n\n  const messageId = await topic.publishMessage({\n    data: messageBuffer,\n    attributes: {\n      type: 'order_created',\n      priority: 'high'\n    }\n  });\n\n  console.log(`Published message ${messageId}`);\n  return messageId;\n}\n```\n\n## Dead Letter Queue\n\n```bash\n# Create DLQ topic\ngcloud pubsub topics create orders-dlq\n\n# Update subscription with DLQ\ngcloud pubsub subscriptions update orders-push \\\n  --dead-letter-topic orders-dlq \\\n  --max-delivery-attempts 5\n```\n\n### Cloud SQL Connection Pattern\n\nConnect Cloud Run to Cloud SQL securely\n\n**When to use**: Need relational database,Migrating existing applications,Complex queries and transactions\n\n```bash\n# Deploy with Cloud SQL connection\ngcloud run deploy my-service \\\n  --add-cloudsql-instances PROJECT:REGION:INSTANCE \\\n  --set-env-vars INSTANCE_CONNECTION_NAME=\"PROJECT:REGION:INSTANCE\" \\\n  --set-env-vars DB_NAME=\"mydb\" \\\n  --set-env-vars DB_USER=\"myuser\"\n```\n\n```javascript\n\u002F\u002F Using Unix socket connection\nconst { Pool } = require('pg');\n\nconst pool = new Pool({\n  user: process.env.DB_USER,\n  password: process.env.DB_PASS,\n  database: process.env.DB_NAME,\n  \u002F\u002F Cloud SQL connector uses Unix socket\n  host: `\u002Fcloudsql\u002F${process.env.INSTANCE_CONNECTION_NAME}`,\n  max: 5,  \u002F\u002F Connection pool size\n  idleTimeoutMillis: 30000,\n  connectionTimeoutMillis: 10000,\n});\n\napp.get('\u002Fapi\u002Fusers', async (req, res) => {\n  const client = await pool.connect();\n  try {\n    const result = await client.query('SELECT * FROM users LIMIT 100');\n    res.json(result.rows);\n  } finally {\n    client.release();\n  }\n});\n```\n\n```python\n# Python with SQLAlchemy\nimport os\nfrom sqlalchemy import create_engine\n\ndef get_engine():\n    instance_connection_name = os.environ[\"INSTANCE_CONNECTION_NAME\"]\n    db_user = os.environ[\"DB_USER\"]\n    db_pass = os.environ[\"DB_PASS\"]\n    db_name = os.environ[\"DB_NAME\"]\n\n    engine = create_engine(\n        f\"postgresql+pg8000:\u002F\u002F{db_user}:{db_pass}@\u002F{db_name}\",\n        connect_args={\n            \"unix_sock\": f\"\u002Fcloudsql\u002F{instance_connection_name}\u002F.s.PGSQL.5432\"\n        },\n        pool_size=5,\n        max_overflow=2,\n        pool_timeout=30,\n        pool_recycle=1800,\n    )\n    return engine\n```\n\n### Best_practices\n\n- Use connection pooling (max 5-10 per instance)\n- Set appropriate idle timeouts\n- Handle connection errors gracefully\n- Consider Cloud SQL Proxy for local development\n\n### Secret Manager Integration\n\nSecurely manage secrets in Cloud Run\n\n**When to use**: API keys, database passwords,Service account keys,Any sensitive configuration\n\n```bash\n# Create secret\necho -n \"my-secret-value\" | gcloud secrets create my-secret --data-file=-\n\n# Mount as environment variable\ngcloud run deploy my-service \\\n  --update-secrets=API_KEY=my-secret:latest\n\n# Mount as file volume\ngcloud run deploy my-service \\\n  --update-secrets=\u002Fsecrets\u002Fapi-key=my-secret:latest\n```\n\n```javascript\n\u002F\u002F Access mounted as environment variable\nconst apiKey = process.env.API_KEY;\n\n\u002F\u002F Access mounted as file\nconst fs = require('fs');\nconst apiKey = fs.readFileSync('\u002Fsecrets\u002Fapi-key', 'utf8');\n\n\u002F\u002F Access via Secret Manager API (when not mounted)\nconst { SecretManagerServiceClient } = require('@google-cloud\u002Fsecret-manager');\nconst client = new SecretManagerServiceClient();\n\nasync function getSecret(name) {\n  const [version] = await client.accessSecretVersion({\n    name: `projects\u002F${projectId}\u002Fsecrets\u002F${name}\u002Fversions\u002Flatest`\n  });\n  return version.payload.data.toString();\n}\n```\n\n## Sharp Edges\n\n### \u002Ftmp Filesystem Counts Against Memory\n\nSeverity: HIGH\n\nSituation: Writing files to \u002Ftmp directory in Cloud Run\n\nSymptoms:\nContainer killed with OOM error.\nMemory usage spikes unexpectedly.\nFile operations cause container restarts.\n\"Container memory limit exceeded\" in logs.\n\nWhy this breaks:\nCloud Run uses an in-memory filesystem for \u002Ftmp. Any files written\nto \u002Ftmp consume memory from your container's allocation.\n\nCommon scenarios:\n- Downloading files temporarily\n- Creating temp processing files\n- Libraries caching to \u002Ftmp\n- Large log buffers\n\nA 512MB container that downloads a 200MB file to \u002Ftmp only has\n~300MB left for the application.\n\nRecommended fix:\n\n## Calculate memory including \u002Ftmp usage\n\n```yaml\n# cloudbuild.yaml\nsteps:\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args:\n      - 'run'\n      - 'deploy'\n      - 'my-service'\n      - '--memory=1Gi'  # Include \u002Ftmp overhead\n      - '--image=gcr.io\u002F$PROJECT_ID\u002Fmy-service'\n```\n\n## Stream instead of buffering\n\n```python\n# BAD - buffers entire file in \u002Ftmp\ndef process_large_file(bucket_name, blob_name):\n    blob = bucket.blob(blob_name)\n    blob.download_to_filename('\u002Ftmp\u002Flarge_file')\n    with open('\u002Ftmp\u002Flarge_file', 'rb') as f:\n        process(f.read())\n\n# GOOD - stream processing\ndef process_large_file(bucket_name, blob_name):\n    blob = bucket.blob(blob_name)\n    with blob.open('rb') as f:\n        for chunk in iter(lambda: f.read(8192), b''):\n            process_chunk(chunk)\n```\n\n## Use Cloud Storage for large files\n\n```python\nfrom google.cloud import storage\n\ndef process_with_gcs(bucket_name, input_blob, output_blob):\n    client = storage.Client()\n    bucket = client.bucket(bucket_name)\n\n    # Process directly to\u002Ffrom GCS\n    input_blob = bucket.blob(input_blob)\n    output_blob = bucket.blob(output_blob)\n\n    with input_blob.open('rb') as reader:\n        with output_blob.open('wb') as writer:\n            for chunk in iter(lambda: reader.read(65536), b''):\n                processed = transform(chunk)\n                writer.write(processed)\n```\n\n## Monitor memory usage\n\n```python\nimport psutil\nimport logging\n\ndef log_memory():\n    memory = psutil.virtual_memory()\n    logging.info(f\"Memory: {memory.percent}% used, \"\n                f\"{memory.available \u002F 1024 \u002F 1024:.0f}MB available\")\n```\n\n### Concurrency=1 Causes Scaling Bottlenecks\n\nSeverity: HIGH\n\nSituation: Setting concurrency to 1 for request isolation\n\nSymptoms:\nAuto-scaling creates many container instances.\nHigh latency during traffic spikes.\nIncreased cold starts.\nHigher costs from more instances.\n\nWhy this breaks:\nSetting concurrency to 1 means each container handles only one\nrequest at a time. During traffic spikes:\n\n- 100 concurrent requests = 100 container instances\n- Each instance has cold start overhead\n- More instances = higher costs\n- Scaling takes time, requests queue up\n\nThis should only be used when:\n- Processing is truly single-threaded\n- Memory-heavy per-request processing\n- Using thread-unsafe libraries\n\nRecommended fix:\n\n## Set appropriate concurrency\n\n```bash\n# For I\u002FO-bound workloads (most web apps)\ngcloud run deploy my-service \\\n  --concurrency=80 \\\n  --max-instances=100\n\n# For CPU-bound workloads\ngcloud run deploy my-service \\\n  --concurrency=4 \\\n  --cpu=2\n\n# Only use 1 when absolutely necessary\ngcloud run deploy my-service \\\n  --concurrency=1 \\\n  --max-instances=1000  # Be prepared for many instances\n```\n\n## Node.js - use async properly\n\n```javascript\n\u002F\u002F With high concurrency, ensure async operations\nconst express = require('express');\nconst app = express();\n\napp.get('\u002Fapi\u002Fdata', async (req, res) => {\n  \u002F\u002F All I\u002FO should be async\n  const data = await fetchFromDatabase();\n  const enriched = await enrichData(data);\n  res.json(enriched);\n});\n\n\u002F\u002F Concurrency 80+ is safe for async I\u002FO workloads\n```\n\n## Python - use async framework\n\n```python\nfrom fastapi import FastAPI\nimport asyncio\nimport httpx\n\napp = FastAPI()\n\n@app.get(\"\u002Fapi\u002Fdata\")\nasync def get_data():\n    # Async I\u002FO allows high concurrency\n    async with httpx.AsyncClient() as client:\n        response = await client.get(\"https:\u002F\u002Fapi.example.com\u002Fdata\")\n        return response.json()\n\n# Concurrency 80+ safe with async framework\n```\n\n## Calculate concurrency\n\n```\nconcurrency = memory_limit \u002F per_request_memory\n\nExample:\n- 512MB container\n- 20MB per request overhead\n- Safe concurrency: ~25\n```\n\n### CPU Throttled When Not Handling Requests\n\nSeverity: HIGH\n\nSituation: Running background tasks or processing between requests\n\nSymptoms:\nBackground tasks run extremely slowly.\nScheduled work doesn't complete.\nMetrics collection fails.\nConnection keep-alive breaks.\n\nWhy this breaks:\nBy default, Cloud Run throttles CPU to near-zero when not actively\nhandling a request. This is \"CPU only during requests\" mode.\n\nAffected operations:\n- Background threads\n- Connection pool maintenance\n- Metrics\u002Ftelemetry emission\n- Scheduled tasks within container\n- Cleanup operations after response\n\nRecommended fix:\n\n## Enable CPU always allocated\n\n```bash\n# CPU allocated even outside requests\ngcloud run deploy my-service \\\n  --cpu-throttling=false \\\n  --min-instances=1\n\n# Note: This increases costs but enables background work\n```\n\n## Use startup CPU boost for initialization\n\n```bash\n# Boost CPU during cold start only\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --cpu-throttling=true  # Default, throttle after request\n```\n\n## Move background work to Cloud Tasks\n\n```python\nfrom google.cloud import tasks_v2\nimport json\n\ndef create_background_task(payload):\n    client = tasks_v2.CloudTasksClient()\n    parent = client.queue_path(\n        \"my-project\", \"us-central1\", \"my-queue\"\n    )\n\n    task = {\n        \"http_request\": {\n            \"http_method\": tasks_v2.HttpMethod.POST,\n            \"url\": \"https:\u002F\u002Fmy-service.run.app\u002Fprocess\",\n            \"body\": json.dumps(payload).encode(),\n            \"headers\": {\"Content-Type\": \"application\u002Fjson\"}\n        }\n    }\n\n    client.create_task(parent=parent, task=task)\n\n# Handle response immediately, background via Cloud Tasks\n@app.post(\"\u002Fapi\u002Forder\")\nasync def create_order(order: Order):\n    order_id = await save_order(order)\n\n    # Queue background processing\n    create_background_task({\"order_id\": order_id})\n\n    return {\"order_id\": order_id, \"status\": \"processing\"}\n```\n\n## Use Pub\u002FSub for async processing\n\n```yaml\n# Move heavy processing to separate service\nsteps:\n  # Main service - responds quickly\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'api-service',\n           '--cpu-throttling=true']\n\n  # Worker service - processes messages\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'worker-service',\n           '--cpu-throttling=false',\n           '--min-instances=1']\n```\n\n### VPC Connector 10-Minute Idle Timeout\n\nSeverity: MEDIUM\n\nSituation: Cloud Run service connecting to VPC resources\n\nSymptoms:\nConnection errors after period of inactivity.\n\"Connection reset\" or \"Connection refused\" errors.\nSporadic failures to VPC resources.\nDatabase connections drop unexpectedly.\n\nWhy this breaks:\nCloud Run's VPC connector has a 10-minute idle timeout on connections.\nIf a connection is idle for 10 minutes, it's silently closed.\n\nAffects:\n- Database connection pools\n- Redis connections\n- Internal API connections\n- Any persistent VPC connection\n\nRecommended fix:\n\n## Configure connection pool with keep-alive\n\n```python\n# SQLAlchemy with connection recycling\nfrom sqlalchemy import create_engine\n\nengine = create_engine(\n    DATABASE_URL,\n    pool_size=5,\n    max_overflow=2,\n    pool_recycle=300,  # Recycle connections every 5 minutes\n    pool_pre_ping=True  # Validate connection before use\n)\n```\n\n## TCP keep-alive for custom connections\n\n```python\nimport socket\n\nsock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)\nsock.setsockopt(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPIDLE, 60)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPINTVL, 60)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPCNT, 5)\n```\n\n## Redis with connection validation\n\n```python\nimport redis\n\npool = redis.ConnectionPool(\n    host=REDIS_HOST,\n    port=6379,\n    socket_keepalive=True,\n    socket_keepalive_options={\n        socket.TCP_KEEPIDLE: 60,\n        socket.TCP_KEEPINTVL: 60,\n        socket.TCP_KEEPCNT: 5\n    },\n    health_check_interval=30\n)\nclient = redis.Redis(connection_pool=pool)\n```\n\n## Use Cloud SQL Proxy sidecar\n\n```yaml\n# Use Cloud SQL connector which handles reconnection\n# requirements.txt\ncloud-sql-python-connector[pg8000]\n```\n\n```python\nfrom google.cloud.sql.connector import Connector\nimport sqlalchemy\n\nconnector = Connector()\n\ndef getconn():\n    return connector.connect(\n        \"project:region:instance\",\n        \"pg8000\",\n        user=\"user\",\n        password=\"password\",\n        db=\"database\"\n    )\n\nengine = sqlalchemy.create_engine(\n    \"postgresql+pg8000:\u002F\u002F\",\n    creator=getconn\n)\n```\n\n### Container Startup Timeout (4 minutes max)\n\nSeverity: HIGH\n\nSituation: Deploying containers with slow initialization\n\nSymptoms:\nDeployment fails with \"Container failed to start\".\nService never becomes healthy.\n\"Revision failed to become ready\" errors.\nWorks locally but fails on Cloud Run.\n\nWhy this breaks:\nCloud Run expects your container to start listening on PORT within\n4 minutes (240 seconds). If it doesn't, the instance is killed.\n\nCommon causes:\n- Heavy framework initialization (ML models, etc.)\n- Waiting for external dependencies at startup\n- Large dependency loading\n- Database migrations on startup\n\nRecommended fix:\n\n## Enable startup CPU boost\n\n```bash\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --startup-cpu-boost\n```\n\n## Lazy initialization\n\n```python\nfrom functools import lru_cache\nfrom fastapi import FastAPI\n\napp = FastAPI()\n\n# Don't load at import time\nmodel = None\n\n@lru_cache()\ndef get_model():\n    global model\n    if model is None:\n        # Load on first request, not at startup\n        model = load_heavy_model()\n    return model\n\n@app.get(\"\u002Fpredict\")\nasync def predict(data: dict):\n    model = get_model()  # Loads on first call only\n    return model.predict(data)\n\n# Startup is fast - model loads on first request\n```\n\n## Start listening immediately\n\n```python\nimport asyncio\nfrom fastapi import FastAPI\nimport uvicorn\n\napp = FastAPI()\n\n# Global state for async initialization\ninitialized = asyncio.Event()\n\n@app.on_event(\"startup\")\nasync def startup():\n    # Start background initialization\n    asyncio.create_task(async_init())\n\nasync def async_init():\n    # Heavy initialization happens after server starts\n    await load_models()\n    await warm_up_connections()\n    initialized.set()\n\n@app.get(\"\u002Fready\")\nasync def ready():\n    if not initialized.is_set():\n        raise HTTPException(503, \"Still initializing\")\n    return {\"status\": \"ready\"}\n\n@app.get(\"\u002Fhealth\")\nasync def health():\n    # Always respond - health check passes\n    return {\"status\": \"healthy\"}\n```\n\n## Use multi-stage builds\n\n```dockerfile\n# Build stage - slow\nFROM python:3.11 as builder\nWORKDIR \u002Fapp\nCOPY requirements.txt .\nRUN pip wheel --no-cache-dir --wheel-dir \u002Fwheels -r requirements.txt\n\n# Runtime stage - fast startup\nFROM python:3.11-slim\nWORKDIR \u002Fapp\nCOPY --from=builder \u002Fwheels \u002Fwheels\nRUN pip install --no-cache \u002Fwheels\u002F* && rm -rf \u002Fwheels\nCOPY . .\nCMD [\"uvicorn\", \"main:app\", \"--host\", \"0.0.0.0\", \"--port\", \"8080\"]\n```\n\n## Run migrations separately\n\n```bash\n# Don't migrate on startup - use Cloud Build\nsteps:\n  # Run migrations first\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    entrypoint: 'bash'\n    args:\n      - '-c'\n      - |\n        gcloud run jobs execute migrate-job --wait\n\n  # Then deploy\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'my-service', ...]\n```\n\n### Second Generation Execution Environment Differences\n\nSeverity: MEDIUM\n\nSituation: Migrating to or using Cloud Run second-gen execution environment\n\nSymptoms:\nNetwork behavior changes.\nDifferent syscall support.\nFile system behavior differences.\nContainer behaves differently than in first-gen.\n\nWhy this breaks:\nCloud Run's second-generation execution environment uses a different\nsandbox (gVisor) with different characteristics:\n\n- More Linux syscalls supported\n- Full \u002Fproc and \u002Fsys access\n- Different network stack\n- No automatic HTTPS redirect\n- Different tmp filesystem behavior\n\nRecommended fix:\n\n## Explicitly set execution environment\n\n```bash\n# First generation (legacy)\ngcloud run deploy my-service \\\n  --execution-environment=gen1\n\n# Second generation (recommended for most)\ngcloud run deploy my-service \\\n  --execution-environment=gen2\n```\n\n## Handle network differences\n\n```python\n# Second-gen doesn't auto-redirect HTTP to HTTPS\nfrom fastapi import FastAPI, Request\nfrom fastapi.responses import RedirectResponse\n\napp = FastAPI()\n\n@app.middleware(\"http\")\nasync def redirect_https(request: Request, call_next):\n    # Check X-Forwarded-Proto header\n    if request.headers.get(\"X-Forwarded-Proto\") == \"http\":\n        url = request.url.replace(scheme=\"https\")\n        return RedirectResponse(url, status_code=301)\n    return await call_next(request)\n```\n\n## GPU access (second-gen only)\n\n```bash\n# GPUs only available in second-gen\ngcloud run deploy ml-service \\\n  --execution-environment=gen2 \\\n  --gpu=1 \\\n  --gpu-type=nvidia-l4\n```\n\n## Check execution environment\n\n```python\nimport os\n\ndef get_execution_environment():\n    # Second-gen has different \u002Fproc structure\n    try:\n        with open('\u002Fproc\u002Fversion', 'r') as f:\n            version = f.read()\n            if 'gVisor' in version:\n                return 'gen2'\n    except:\n        pass\n    return 'gen1'\n```\n\n### Request Timeout Configuration Mismatch\n\nSeverity: MEDIUM\n\nSituation: Long-running requests or background processing\n\nSymptoms:\nRequests terminated before completion.\n504 Gateway Timeout errors.\nProcessing stops unexpectedly.\nInconsistent timeout behavior.\n\nWhy this breaks:\nCloud Run has multiple timeout configurations that must align:\n- Request timeout (default 300s, max 3600s for HTTP, 60m for gRPC)\n- Client timeout\n- Downstream service timeouts\n- Load balancer timeout (for external access)\n\nRecommended fix:\n\n## Set consistent timeouts\n\n```bash\n# Increase request timeout (max 3600s for HTTP)\ngcloud run deploy my-service \\\n  --timeout=900  # 15 minutes\n```\n\n## Handle long-running with webhooks\n\n```python\nfrom fastapi import FastAPI, BackgroundTasks\nimport httpx\n\napp = FastAPI()\n\n@app.post(\"\u002Fprocess\")\nasync def process(data: dict, background_tasks: BackgroundTasks):\n    task_id = create_task_id()\n\n    # Start background processing\n    background_tasks.add_task(\n        long_running_process,\n        task_id,\n        data,\n        data.get(\"callback_url\")\n    )\n\n    # Return immediately\n    return {\"task_id\": task_id, \"status\": \"processing\"}\n\nasync def long_running_process(task_id, data, callback_url):\n    result = await heavy_computation(data)\n\n    # Callback when done\n    if callback_url:\n        async with httpx.AsyncClient() as client:\n            await client.post(callback_url, json={\n                \"task_id\": task_id,\n                \"result\": result\n            })\n```\n\n## Use Cloud Tasks for reliable long-running\n\n```python\nfrom google.cloud import tasks_v2\n\ndef create_long_running_task(data):\n    client = tasks_v2.CloudTasksClient()\n    parent = client.queue_path(PROJECT, REGION, \"long-tasks\")\n\n    task = {\n        \"http_request\": {\n            \"http_method\": tasks_v2.HttpMethod.POST,\n            \"url\": \"https:\u002F\u002Fworker.run.app\u002Fprocess\",\n            \"body\": json.dumps(data).encode(),\n            \"headers\": {\"Content-Type\": \"application\u002Fjson\"}\n        },\n        \"dispatch_deadline\": {\"seconds\": 1800}  # 30 min\n    }\n\n    return client.create_task(parent=parent, task=task)\n```\n\n## Streaming for long responses\n\n```python\nfrom fastapi import FastAPI\nfrom fastapi.responses import StreamingResponse\n\n@app.get(\"\u002Flarge-report\")\nasync def large_report():\n    async def generate():\n        for chunk in process_large_data():\n            yield chunk\n\n    return StreamingResponse(generate(), media_type=\"text\u002Fplain\")\n```\n\n## Validation Checks\n\n### Hardcoded GCP Credentials\n\nSeverity: ERROR\n\nGCP credentials must never be hardcoded in source code\n\nMessage: Hardcoded GCP service account credentials. Use Secret Manager or Workload Identity.\n\n### GCP API Key in Source Code\n\nSeverity: ERROR\n\nAPI keys should use Secret Manager\n\nMessage: Hardcoded GCP API key. Use Secret Manager.\n\n### Credentials JSON File in Repository\n\nSeverity: ERROR\n\nService account JSON files should not be in source control\n\nMessage: Credentials file detected. Add to .gitignore and use Secret Manager.\n\n### Running as Root User\n\nSeverity: WARNING\n\nContainers should not run as root for security\n\nMessage: Dockerfile runs as root. Add USER directive for security.\n\n### Missing Health Check in Dockerfile\n\nSeverity: INFO\n\nCloud Run uses HTTP health checks, Dockerfile HEALTHCHECK is optional\n\nMessage: No HEALTHCHECK in Dockerfile. Cloud Run uses its own health checks.\n\n### Hardcoded Port in Application\n\nSeverity: WARNING\n\nPort should come from PORT environment variable\n\nMessage: Hardcoded port. Use PORT environment variable for Cloud Run.\n\n### Large File Writes to \u002Ftmp\n\nSeverity: WARNING\n\n\u002Ftmp uses container memory, large writes can cause OOM\n\nMessage: \u002Ftmp writes consume memory. Consider Cloud Storage for large files.\n\n### Synchronous File Operations\n\nSeverity: WARNING\n\nSync file ops block the event loop in async apps\n\nMessage: Synchronous file operations. Use async versions for better concurrency.\n\n### Global Mutable State\n\nSeverity: WARNING\n\nGlobal state issues with concurrent requests\n\nMessage: Global mutable state may cause issues with concurrent requests.\n\n### Thread-Unsafe Singleton Pattern\n\nSeverity: WARNING\n\nSingletons need thread safety for concurrency > 1\n\nMessage: Singleton pattern - ensure thread safety if using concurrency > 1.\n\n## Collaboration\n\n### Delegation Triggers\n\n- user needs AWS serverless -> aws-serverless (Lambda, API Gateway, SAM)\n- user needs Azure containers -> azure-functions (Azure Container Apps, Functions)\n- user needs database design -> postgres-wizard (Cloud SQL design, AlloyDB)\n- user needs authentication -> auth-specialist (Firebase Auth, Identity Platform)\n- user needs AI integration -> llm-architect (Vertex AI, Cloud Run + LLM)\n- user needs workflow orchestration -> workflow-automation (Cloud Workflows, Eventarc)\n\n## When to Use\nUse this skill when the request clearly matches the capabilities and patterns described above.\n\n## Limitations\n- Use this skill only when the task clearly matches the scope described above.\n- Do not treat the output as a substitute for environment-specific validation, testing, or expert review.\n- Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.\n","# GCP Cloud Run\n\nSpecialized skill for building production-ready serverless applications on GCP.\nCovers Cloud Run services (containerized), Cloud Run Functions (event-driven),\ncold start optimization, and event-driven architecture with Pub\u002FSub.\n\n## Principles\n\n- Cloud Run for containers, Functions for simple event handlers\n- Optimize for cold starts with startup CPU boost and min instances\n- Set concurrency based on workload (start with 8, adjust)\n- Memory includes \u002Ftmp filesystem - plan accordingly\n- Use VPC Connector only when needed (adds latency)\n- Containers should start fast and be stateless\n- Handle signals gracefully for clean shutdown\n\n## Patterns\n\n### Cloud Run Service Pattern\n\nContainerized web service on Cloud Run\n\n**When to use**: Web applications and APIs,Need any runtime or library,Complex services with multiple endpoints,Stateless containerized workloads\n\n```dockerfile\n# Dockerfile - Multi-stage build for smaller image\nFROM node:20-slim AS builder\nWORKDIR \u002Fapp\nCOPY package*.json .\u002F\nRUN npm ci --only=production\n\nFROM node:20-slim\nWORKDIR \u002Fapp\n\n# Copy only production dependencies\nCOPY --from=builder \u002Fapp\u002Fnode_modules .\u002Fnode_modules\nCOPY src .\u002Fsrc\nCOPY package.json .\u002F\n\n# Cloud Run uses PORT env variable\nENV PORT=8080\nEXPOSE 8080\n\n# Run as non-root user\nUSER node\n\nCMD [\"node\", \"src\u002Findex.js\"]\n```\n\n```javascript\n\u002F\u002F src\u002Findex.js\nconst express = require('express');\nconst app = express();\n\napp.use(express.json());\n\n\u002F\u002F Health check endpoint\napp.get('\u002Fhealth', (req, res) => {\n  res.status(200).send('OK');\n});\n\n\u002F\u002F API routes\napp.get('\u002Fapi\u002Fitems\u002F:id', async (req, res) => {\n  try {\n    const item = await getItem(req.params.id);\n    res.json(item);\n  } catch (error) {\n    console.error('Error:', error);\n    res.status(500).json({ error: 'Internal server error' });\n  }\n});\n\n\u002F\u002F Graceful shutdown\nprocess.on('SIGTERM', () => {\n  console.log('SIGTERM received, shutting down gracefully');\n  server.close(() => {\n    console.log('Server closed');\n    process.exit(0);\n  });\n});\n\nconst PORT = process.env.PORT || 8080;\nconst server = app.listen(PORT, () => {\n  console.log(`Server listening on port ${PORT}`);\n});\n```\n\n```yaml\n# cloudbuild.yaml\nsteps:\n  # Build the container image\n  - name: 'gcr.io\u002Fcloud-builders\u002Fdocker'\n    args: ['build', '-t', 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA', '.']\n\n  # Push the container image\n  - name: 'gcr.io\u002Fcloud-builders\u002Fdocker'\n    args: ['push', 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA']\n\n  # Deploy to Cloud Run\n  - name: 'gcr.io\u002Fgoogle.com\u002Fcloudsdktool\u002Fcloud-sdk'\n    entrypoint: gcloud\n    args:\n      - 'run'\n      - 'deploy'\n      - 'my-service'\n      - '--image=gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA'\n      - '--region=us-central1'\n      - '--platform=managed'\n      - '--allow-unauthenticated'\n      - '--memory=512Mi'\n      - '--cpu=1'\n      - '--min-instances=1'\n      - '--max-instances=100'\n      - '--concurrency=80'\n      - '--cpu-boost'\n\nimages:\n  - 'gcr.io\u002F$PROJECT_ID\u002Fmy-service:$COMMIT_SHA'\n```\n\n### Structure\n\nproject\u002F\n├── Dockerfile\n├── .dockerignore\n├── src\u002F\n│   ├── index.js\n│   └── routes\u002F\n├── package.json\n└── cloudbuild.yaml\n\n### Gcloud_deploy\n\n# Direct gcloud deployment\ngcloud run deploy my-service \\\n  --source . \\\n  --region us-central1 \\\n  --allow-unauthenticated \\\n  --memory 512Mi \\\n  --cpu 1 \\\n  --min-instances 1 \\\n  --max-instances 100 \\\n  --concurrency 80 \\\n  --cpu-boost\n\n### Cloud Run Functions Pattern\n\nEvent-driven functions (formerly Cloud Functions)\n\n**When to use**: Simple event handlers,Pub\u002FSub message processing,Cloud Storage triggers,HTTP webhooks\n\n```javascript\n\u002F\u002F HTTP Function\n\u002F\u002F index.js\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.http('helloHttp', (req, res) => {\n  const name = req.query.name || req.body.name || 'World';\n  res.send(`Hello, ${name}!`);\n});\n```\n\n```javascript\n\u002F\u002F Pub\u002FSub Function\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.cloudEvent('processPubSub', (cloudEvent) => {\n  \u002F\u002F Decode Pub\u002FSub message\n  const message = cloudEvent.data.message;\n  const data = message.data\n    ? JSON.parse(Buffer.from(message.data, 'base64').toString())\n    : {};\n\n  console.log('Received message:', data);\n\n  \u002F\u002F Process message\n  processMessage(data);\n});\n```\n\n```javascript\n\u002F\u002F Cloud Storage Function\nconst functions = require('@google-cloud\u002Ffunctions-framework');\n\nfunctions.cloudEvent('processStorageEvent', async (cloudEvent) => {\n  const file = cloudEvent.data;\n\n  console.log(`Event: ${cloudEvent.type}`);\n  console.log(`Bucket: ${file.bucket}`);\n  console.log(`File: ${file.name}`);\n\n  if (cloudEvent.type === 'google.cloud.storage.object.v1.finalized') {\n    await processUploadedFile(file.bucket, file.name);\n  }\n});\n```\n\n```bash\n# Deploy HTTP function\ngcloud functions deploy hello-http \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-http \\\n  --allow-unauthenticated \\\n  --region us-central1\n\n# Deploy Pub\u002FSub function\ngcloud functions deploy process-messages \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-topic my-topic \\\n  --region us-central1\n\n# Deploy Cloud Storage function\ngcloud functions deploy process-uploads \\\n  --gen2 \\\n  --runtime nodejs20 \\\n  --trigger-event-filters=\"type=google.cloud.storage.object.v1.finalized\" \\\n  --trigger-event-filters=\"bucket=my-bucket\" \\\n  --region us-central1\n```\n\n### Cold Start Optimization Pattern\n\nMinimize cold start latency for Cloud Run\n\n**When to use**: Latency-sensitive applications,User-facing APIs,High-traffic services\n\n## 1. Enable Startup CPU Boost\n\n```bash\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --region us-central1\n```\n\n## 2. Set Minimum Instances\n\n```bash\ngcloud run deploy my-service \\\n  --min-instances 1 \\\n  --region us-central1\n```\n\n## 3. Optimize Container Image\n\n```dockerfile\n# Use distroless for minimal image\nFROM node:20-slim AS builder\nWORKDIR \u002Fapp\nCOPY package*.json .\u002F\nRUN npm ci --only=production\n\nFROM gcr.io\u002Fdistroless\u002Fnodejs20-debian12\nWORKDIR \u002Fapp\nCOPY --from=builder \u002Fapp\u002Fnode_modules .\u002Fnode_modules\nCOPY src .\u002Fsrc\nCMD [\"src\u002Findex.js\"]\n```\n\n## 4. Lazy Initialize Heavy Dependencies\n\n```javascript\n\u002F\u002F Lazy load heavy libraries\nlet bigQueryClient = null;\n\nfunction getBigQueryClient() {\n  if (!bigQueryClient) {\n    const { BigQuery } = require('@google-cloud\u002Fbigquery');\n    bigQueryClient = new BigQuery();\n  }\n  return bigQueryClient;\n}\n\n\u002F\u002F Only initialize when needed\napp.get('\u002Fapi\u002Fanalytics', async (req, res) => {\n  const client = getBigQueryClient();\n  const results = await client.query({...});\n  res.json(results);\n});\n```\n\n## 5. Increase Memory (More CPU)\n\n```bash\n# Higher memory = more CPU during startup\ngcloud run deploy my-service \\\n  --memory 1Gi \\\n  --cpu 2 \\\n  --region us-central1\n```\n\n### Optimization_impact\n\n- Startup_cpu_boost: 50% faster cold starts\n- Min_instances: Eliminates cold starts for traffic spikes\n- Distroless_image: Smaller attack surface, faster pull\n- Lazy_init: Defers heavy loading to first request\n\n### Concurrency Configuration Pattern\n\nProper concurrency settings for Cloud Run\n\n**When to use**: Need to optimize instance utilization,Handle traffic spikes efficiently,Reduce cold starts\n\n## Understanding Concurrency\n\n```bash\n# Default concurrency is 80\n# Adjust based on your workload\n\n# For I\u002FO-bound workloads (most web apps)\ngcloud run deploy my-service \\\n  --concurrency 80 \\\n  --cpu 1\n\n# For CPU-bound workloads\ngcloud run deploy my-service \\\n  --concurrency 1 \\\n  --cpu 1\n\n# For memory-intensive workloads\ngcloud run deploy my-service \\\n  --concurrency 10 \\\n  --memory 2Gi\n```\n\n## Node.js Concurrency\n\n```javascript\n\u002F\u002F Node.js is single-threaded but handles I\u002FO concurrently\n\u002F\u002F Use async\u002Fawait for all I\u002FO operations\n\n\u002F\u002F GOOD - async I\u002FO\napp.get('\u002Fapi\u002Fdata', async (req, res) => {\n  const [users, products] = await Promise.all([\n    fetchUsers(),\n    fetchProducts()\n  ]);\n  res.json({ users, products });\n});\n\n\u002F\u002F BAD - blocking operation\napp.get('\u002Fapi\u002Fcompute', (req, res) => {\n  const result = heavyCpuOperation(); \u002F\u002F Blocks other requests!\n  res.json(result);\n});\n```\n\n## Python Concurrency with Gunicorn\n\n```dockerfile\nFROM python:3.11-slim\nWORKDIR \u002Fapp\nCOPY requirements.txt .\nRUN pip install --no-cache-dir -r requirements.txt\nCOPY . .\n\n# 4 workers for concurrency\nCMD exec gunicorn --bind :$PORT --workers 4 --threads 2 main:app\n```\n\n```python\n# main.py\nfrom flask import Flask\napp = Flask(__name__)\n\n@app.route('\u002Fapi\u002Fdata')\ndef get_data():\n    return {'status': 'ok'}\n```\n\n### Concurrency_guidelines\n\n- Concurrency=1: Only for CPU-bound or unsafe code\n- Concurrency=8 20: Memory-intensive workloads\n- Concurrency=80: Default, good for I\u002FO-bound\n- Concurrency=250: Maximum, for very lightweight handlers\n\n### Pub\u002FSub Integration Pattern\n\nEvent-driven processing with Cloud Pub\u002FSub\n\n**When to use**: Asynchronous message processing,Decoupled microservices,Event-driven architecture\n\n## Push Subscription to Cloud Run\n\n```bash\n# Create topic\ngcloud pubsub topics create orders\n\n# Create push subscription to Cloud Run\ngcloud pubsub subscriptions create orders-push \\\n  --topic orders \\\n  --push-endpoint https:\u002F\u002Fmy-service-xxx.run.app\u002Fpubsub \\\n  --ack-deadline 600\n```\n\n```javascript\n\u002F\u002F Handle Pub\u002FSub push messages\nconst express = require('express');\nconst app = express();\napp.use(express.json());\n\napp.post('\u002Fpubsub', async (req, res) => {\n  \u002F\u002F Verify the request is from Pub\u002FSub\n  if (!req.body.message) {\n    return res.status(400).send('Invalid Pub\u002FSub message');\n  }\n\n  try {\n    \u002F\u002F Decode message data\n    const message = req.body.message;\n    const data = message.data\n      ? JSON.parse(Buffer.from(message.data, 'base64').toString())\n      : {};\n\n    console.log('Processing order:', data);\n\n    await processOrder(data);\n\n    \u002F\u002F Return 200 to acknowledge\n    res.status(200).send('OK');\n  } catch (error) {\n    console.error('Processing failed:', error);\n    \u002F\u002F Return 500 to trigger retry\n    res.status(500).send('Processing failed');\n  }\n});\n```\n\n## Publishing Messages\n\n```javascript\nconst { PubSub } = require('@google-cloud\u002Fpubsub');\nconst pubsub = new PubSub();\n\nasync function publishOrder(order) {\n  const topic = pubsub.topic('orders');\n  const messageBuffer = Buffer.from(JSON.stringify(order));\n\n  const messageId = await topic.publishMessage({\n    data: messageBuffer,\n    attributes: {\n      type: 'order_created',\n      priority: 'high'\n    }\n  });\n\n  console.log(`Published message ${messageId}`);\n  return messageId;\n}\n```\n\n## Dead Letter Queue\n\n```bash\n# Create DLQ topic\ngcloud pubsub topics create orders-dlq\n\n# Update subscription with DLQ\ngcloud pubsub subscriptions update orders-push \\\n  --dead-letter-topic orders-dlq \\\n  --max-delivery-attempts 5\n```\n\n### Cloud SQL Connection Pattern\n\nConnect Cloud Run to Cloud SQL securely\n\n**When to use**: Need relational database,Migrating existing applications,Complex queries and transactions\n\n```bash\n# Deploy with Cloud SQL connection\ngcloud run deploy my-service \\\n  --add-cloudsql-instances PROJECT:REGION:INSTANCE \\\n  --set-env-vars INSTANCE_CONNECTION_NAME=\"PROJECT:REGION:INSTANCE\" \\\n  --set-env-vars DB_NAME=\"mydb\" \\\n  --set-env-vars DB_USER=\"myuser\"\n```\n\n```javascript\n\u002F\u002F Using Unix socket connection\nconst { Pool } = require('pg');\n\nconst pool = new Pool({\n  user: process.env.DB_USER,\n  password: process.env.DB_PASS,\n  database: process.env.DB_NAME,\n  \u002F\u002F Cloud SQL connector uses Unix socket\n  host: `\u002Fcloudsql\u002F${process.env.INSTANCE_CONNECTION_NAME}`,\n  max: 5,  \u002F\u002F Connection pool size\n  idleTimeoutMillis: 30000,\n  connectionTimeoutMillis: 10000,\n});\n\napp.get('\u002Fapi\u002Fusers', async (req, res) => {\n  const client = await pool.connect();\n  try {\n    const result = await client.query('SELECT * FROM users LIMIT 100');\n    res.json(result.rows);\n  } finally {\n    client.release();\n  }\n});\n```\n\n```python\n# Python with SQLAlchemy\nimport os\nfrom sqlalchemy import create_engine\n\ndef get_engine():\n    instance_connection_name = os.environ[\"INSTANCE_CONNECTION_NAME\"]\n    db_user = os.environ[\"DB_USER\"]\n    db_pass = os.environ[\"DB_PASS\"]\n    db_name = os.environ[\"DB_NAME\"]\n\n    engine = create_engine(\n        f\"postgresql+pg8000:\u002F\u002F{db_user}:{db_pass}@\u002F{db_name}\",\n        connect_args={\n            \"unix_sock\": f\"\u002Fcloudsql\u002F{instance_connection_name}\u002F.s.PGSQL.5432\"\n        },\n        pool_size=5,\n        max_overflow=2,\n        pool_timeout=30,\n        pool_recycle=1800,\n    )\n    return engine\n```\n\n### Best_practices\n\n- Use connection pooling (max 5-10 per instance)\n- Set appropriate idle timeouts\n- Handle connection errors gracefully\n- Consider Cloud SQL Proxy for local development\n\n### Secret Manager Integration\n\nSecurely manage secrets in Cloud Run\n\n**When to use**: API keys, database passwords,Service account keys,Any sensitive configuration\n\n```bash\n# Create secret\necho -n \"my-secret-value\" | gcloud secrets create my-secret --data-file=-\n\n# Mount as environment variable\ngcloud run deploy my-service \\\n  --update-secrets=API_KEY=my-secret:latest\n\n# Mount as file volume\ngcloud run deploy my-service \\\n  --update-secrets=\u002Fsecrets\u002Fapi-key=my-secret:latest\n```\n\n```javascript\n\u002F\u002F Access mounted as environment variable\nconst apiKey = process.env.API_KEY;\n\n\u002F\u002F Access mounted as file\nconst fs = require('fs');\nconst apiKey = fs.readFileSync('\u002Fsecrets\u002Fapi-key', 'utf8');\n\n\u002F\u002F Access via Secret Manager API (when not mounted)\nconst { SecretManagerServiceClient } = require('@google-cloud\u002Fsecret-manager');\nconst client = new SecretManagerServiceClient();\n\nasync function getSecret(name) {\n  const [version] = await client.accessSecretVersion({\n    name: `projects\u002F${projectId}\u002Fsecrets\u002F${name}\u002Fversions\u002Flatest`\n  });\n  return version.payload.data.toString();\n}\n```\n\n## Sharp Edges\n\n### \u002Ftmp Filesystem Counts Against Memory\n\nSeverity: HIGH\n\nSituation: Writing files to \u002Ftmp directory in Cloud Run\n\nSymptoms:\nContainer killed with OOM error.\nMemory usage spikes unexpectedly.\nFile operations cause container restarts.\n\"Container memory limit exceeded\" in logs.\n\nWhy this breaks:\nCloud Run uses an in-memory filesystem for \u002Ftmp. Any files written\nto \u002Ftmp consume memory from your container's allocation.\n\nCommon scenarios:\n- Downloading files temporarily\n- Creating temp processing files\n- Libraries caching to \u002Ftmp\n- Large log buffers\n\nA 512MB container that downloads a 200MB file to \u002Ftmp only has\n~300MB left for the application.\n\nRecommended fix:\n\n## Calculate memory including \u002Ftmp usage\n\n```yaml\n# cloudbuild.yaml\nsteps:\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args:\n      - 'run'\n      - 'deploy'\n      - 'my-service'\n      - '--memory=1Gi'  # Include \u002Ftmp overhead\n      - '--image=gcr.io\u002F$PROJECT_ID\u002Fmy-service'\n```\n\n## Stream instead of buffering\n\n```python\n# BAD - buffers entire file in \u002Ftmp\ndef process_large_file(bucket_name, blob_name):\n    blob = bucket.blob(blob_name)\n    blob.download_to_filename('\u002Ftmp\u002Flarge_file')\n    with open('\u002Ftmp\u002Flarge_file', 'rb') as f:\n        process(f.read())\n\n# GOOD - stream processing\ndef process_large_file(bucket_name, blob_name):\n    blob = bucket.blob(blob_name)\n    with blob.open('rb') as f:\n        for chunk in iter(lambda: f.read(8192), b''):\n            process_chunk(chunk)\n```\n\n## Use Cloud Storage for large files\n\n```python\nfrom google.cloud import storage\n\ndef process_with_gcs(bucket_name, input_blob, output_blob):\n    client = storage.Client()\n    bucket = client.bucket(bucket_name)\n\n    # Process directly to\u002Ffrom GCS\n    input_blob = bucket.blob(input_blob)\n    output_blob = bucket.blob(output_blob)\n\n    with input_blob.open('rb') as reader:\n        with output_blob.open('wb') as writer:\n            for chunk in iter(lambda: reader.read(65536), b''):\n                processed = transform(chunk)\n                writer.write(processed)\n```\n\n## Monitor memory usage\n\n```python\nimport psutil\nimport logging\n\ndef log_memory():\n    memory = psutil.virtual_memory()\n    logging.info(f\"Memory: {memory.percent}% used, \"\n                f\"{memory.available \u002F 1024 \u002F 1024:.0f}MB available\")\n```\n\n### Concurrency=1 Causes Scaling Bottlenecks\n\nSeverity: HIGH\n\nSituation: Setting concurrency to 1 for request isolation\n\nSymptoms:\nAuto-scaling creates many container instances.\nHigh latency during traffic spikes.\nIncreased cold starts.\nHigher costs from more instances.\n\nWhy this breaks:\nSetting concurrency to 1 means each container handles only one\nrequest at a time. During traffic spikes:\n\n- 100 concurrent requests = 100 container instances\n- Each instance has cold start overhead\n- More instances = higher costs\n- Scaling takes time, requests queue up\n\nThis should only be used when:\n- Processing is truly single-threaded\n- Memory-heavy per-request processing\n- Using thread-unsafe libraries\n\nRecommended fix:\n\n## Set appropriate concurrency\n\n```bash\n# For I\u002FO-bound workloads (most web apps)\ngcloud run deploy my-service \\\n  --concurrency=80 \\\n  --max-instances=100\n\n# For CPU-bound workloads\ngcloud run deploy my-service \\\n  --concurrency=4 \\\n  --cpu=2\n\n# Only use 1 when absolutely necessary\ngcloud run deploy my-service \\\n  --concurrency=1 \\\n  --max-instances=1000  # Be prepared for many instances\n```\n\n## Node.js - use async properly\n\n```javascript\n\u002F\u002F With high concurrency, ensure async operations\nconst express = require('express');\nconst app = express();\n\napp.get('\u002Fapi\u002Fdata', async (req, res) => {\n  \u002F\u002F All I\u002FO should be async\n  const data = await fetchFromDatabase();\n  const enriched = await enrichData(data);\n  res.json(enriched);\n});\n\n\u002F\u002F Concurrency 80+ is safe for async I\u002FO workloads\n```\n\n## Python - use async framework\n\n```python\nfrom fastapi import FastAPI\nimport asyncio\nimport httpx\n\napp = FastAPI()\n\n@app.get(\"\u002Fapi\u002Fdata\")\nasync def get_data():\n    # Async I\u002FO allows high concurrency\n    async with httpx.AsyncClient() as client:\n        response = await client.get(\"https:\u002F\u002Fapi.example.com\u002Fdata\")\n        return response.json()\n\n# Concurrency 80+ safe with async framework\n```\n\n## Calculate concurrency\n\n```\nconcurrency = memory_limit \u002F per_request_memory\n\nExample:\n- 512MB container\n- 20MB per request overhead\n- Safe concurrency: ~25\n```\n\n### CPU Throttled When Not Handling Requests\n\nSeverity: HIGH\n\nSituation: Running background tasks or processing between requests\n\nSymptoms:\nBackground tasks run extremely slowly.\nScheduled work doesn't complete.\nMetrics collection fails.\nConnection keep-alive breaks.\n\nWhy this breaks:\nBy default, Cloud Run throttles CPU to near-zero when not actively\nhandling a request. This is \"CPU only during requests\" mode.\n\nAffected operations:\n- Background threads\n- Connection pool maintenance\n- Metrics\u002Ftelemetry emission\n- Scheduled tasks within container\n- Cleanup operations after response\n\nRecommended fix:\n\n## Enable CPU always allocated\n\n```bash\n# CPU allocated even outside requests\ngcloud run deploy my-service \\\n  --cpu-throttling=false \\\n  --min-instances=1\n\n# Note: This increases costs but enables background work\n```\n\n## Use startup CPU boost for initialization\n\n```bash\n# Boost CPU during cold start only\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --cpu-throttling=true  # Default, throttle after request\n```\n\n## Move background work to Cloud Tasks\n\n```python\nfrom google.cloud import tasks_v2\nimport json\n\ndef create_background_task(payload):\n    client = tasks_v2.CloudTasksClient()\n    parent = client.queue_path(\n        \"my-project\", \"us-central1\", \"my-queue\"\n    )\n\n    task = {\n        \"http_request\": {\n            \"http_method\": tasks_v2.HttpMethod.POST,\n            \"url\": \"https:\u002F\u002Fmy-service.run.app\u002Fprocess\",\n            \"body\": json.dumps(payload).encode(),\n            \"headers\": {\"Content-Type\": \"application\u002Fjson\"}\n        }\n    }\n\n    client.create_task(parent=parent, task=task)\n\n# Handle response immediately, background via Cloud Tasks\n@app.post(\"\u002Fapi\u002Forder\")\nasync def create_order(order: Order):\n    order_id = await save_order(order)\n\n    # Queue background processing\n    create_background_task({\"order_id\": order_id})\n\n    return {\"order_id\": order_id, \"status\": \"processing\"}\n```\n\n## Use Pub\u002FSub for async processing\n\n```yaml\n# Move heavy processing to separate service\nsteps:\n  # Main service - responds quickly\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'api-service',\n           '--cpu-throttling=true']\n\n  # Worker service - processes messages\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'worker-service',\n           '--cpu-throttling=false',\n           '--min-instances=1']\n```\n\n### VPC Connector 10-Minute Idle Timeout\n\nSeverity: MEDIUM\n\nSituation: Cloud Run service connecting to VPC resources\n\nSymptoms:\nConnection errors after period of inactivity.\n\"Connection reset\" or \"Connection refused\" errors.\nSporadic failures to VPC resources.\nDatabase connections drop unexpectedly.\n\nWhy this breaks:\nCloud Run's VPC connector has a 10-minute idle timeout on connections.\nIf a connection is idle for 10 minutes, it's silently closed.\n\nAffects:\n- Database connection pools\n- Redis connections\n- Internal API connections\n- Any persistent VPC connection\n\nRecommended fix:\n\n## Configure connection pool with keep-alive\n\n```python\n# SQLAlchemy with connection recycling\nfrom sqlalchemy import create_engine\n\nengine = create_engine(\n    DATABASE_URL,\n    pool_size=5,\n    max_overflow=2,\n    pool_recycle=300,  # Recycle connections every 5 minutes\n    pool_pre_ping=True  # Validate connection before use\n)\n```\n\n## TCP keep-alive for custom connections\n\n```python\nimport socket\n\nsock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)\nsock.setsockopt(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPIDLE, 60)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPINTVL, 60)\nsock.setsockopt(socket.IPPROTO_TCP, socket.TCP_KEEPCNT, 5)\n```\n\n## Redis with connection validation\n\n```python\nimport redis\n\npool = redis.ConnectionPool(\n    host=REDIS_HOST,\n    port=6379,\n    socket_keepalive=True,\n    socket_keepalive_options={\n        socket.TCP_KEEPIDLE: 60,\n        socket.TCP_KEEPINTVL: 60,\n        socket.TCP_KEEPCNT: 5\n    },\n    health_check_interval=30\n)\nclient = redis.Redis(connection_pool=pool)\n```\n\n## Use Cloud SQL Proxy sidecar\n\n```yaml\n# Use Cloud SQL connector which handles reconnection\n# requirements.txt\ncloud-sql-python-connector[pg8000]\n```\n\n```python\nfrom google.cloud.sql.connector import Connector\nimport sqlalchemy\n\nconnector = Connector()\n\ndef getconn():\n    return connector.connect(\n        \"project:region:instance\",\n        \"pg8000\",\n        user=\"user\",\n        password=\"password\",\n        db=\"database\"\n    )\n\nengine = sqlalchemy.create_engine(\n    \"postgresql+pg8000:\u002F\u002F\",\n    creator=getconn\n)\n```\n\n### Container Startup Timeout (4 minutes max)\n\nSeverity: HIGH\n\nSituation: Deploying containers with slow initialization\n\nSymptoms:\nDeployment fails with \"Container failed to start\".\nService never becomes healthy.\n\"Revision failed to become ready\" errors.\nWorks locally but fails on Cloud Run.\n\nWhy this breaks:\nCloud Run expects your container to start listening on PORT within\n4 minutes (240 seconds). If it doesn't, the instance is killed.\n\nCommon causes:\n- Heavy framework initialization (ML models, etc.)\n- Waiting for external dependencies at startup\n- Large dependency loading\n- Database migrations on startup\n\nRecommended fix:\n\n## Enable startup CPU boost\n\n```bash\ngcloud run deploy my-service \\\n  --cpu-boost \\\n  --startup-cpu-boost\n```\n\n## Lazy initialization\n\n```python\nfrom functools import lru_cache\nfrom fastapi import FastAPI\n\napp = FastAPI()\n\n# Don't load at import time\nmodel = None\n\n@lru_cache()\ndef get_model():\n    global model\n    if model is None:\n        # Load on first request, not at startup\n        model = load_heavy_model()\n    return model\n\n@app.get(\"\u002Fpredict\")\nasync def predict(data: dict):\n    model = get_model()  # Loads on first call only\n    return model.predict(data)\n\n# Startup is fast - model loads on first request\n```\n\n## Start listening immediately\n\n```python\nimport asyncio\nfrom fastapi import FastAPI\nimport uvicorn\n\napp = FastAPI()\n\n# Global state for async initialization\ninitialized = asyncio.Event()\n\n@app.on_event(\"startup\")\nasync def startup():\n    # Start background initialization\n    asyncio.create_task(async_init())\n\nasync def async_init():\n    # Heavy initialization happens after server starts\n    await load_models()\n    await warm_up_connections()\n    initialized.set()\n\n@app.get(\"\u002Fready\")\nasync def ready():\n    if not initialized.is_set():\n        raise HTTPException(503, \"Still initializing\")\n    return {\"status\": \"ready\"}\n\n@app.get(\"\u002Fhealth\")\nasync def health():\n    # Always respond - health check passes\n    return {\"status\": \"healthy\"}\n```\n\n## Use multi-stage builds\n\n```dockerfile\n# Build stage - slow\nFROM python:3.11 as builder\nWORKDIR \u002Fapp\nCOPY requirements.txt .\nRUN pip wheel --no-cache-dir --wheel-dir \u002Fwheels -r requirements.txt\n\n# Runtime stage - fast startup\nFROM python:3.11-slim\nWORKDIR \u002Fapp\nCOPY --from=builder \u002Fwheels \u002Fwheels\nRUN pip install --no-cache \u002Fwheels\u002F* && rm -rf \u002Fwheels\nCOPY . .\nCMD [\"uvicorn\", \"main:app\", \"--host\", \"0.0.0.0\", \"--port\", \"8080\"]\n```\n\n## Run migrations separately\n\n```bash\n# Don't migrate on startup - use Cloud Build\nsteps:\n  # Run migrations first\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    entrypoint: 'bash'\n    args:\n      - '-c'\n      - |\n        gcloud run jobs execute migrate-job --wait\n\n  # Then deploy\n  - name: 'gcr.io\u002Fcloud-builders\u002Fgcloud'\n    args: ['run', 'deploy', 'my-service', ...]\n```\n\n### Second Generation Execution Environment Differences\n\nSeverity: MEDIUM\n\nSituation: Migrating to or using Cloud Run second-gen execution environment\n\nSymptoms:\nNetwork behavior changes.\nDifferent syscall support.\nFile system behavior differences.\nContainer behaves differently than in first-gen.\n\nWhy this breaks:\nCloud Run's second-generation execution environment uses a different\nsandbox (gVisor) with different characteristics:\n\n- More Linux syscalls supported\n- Full \u002Fproc and \u002Fsys access\n- Different network stack\n- No automatic HTTPS redirect\n- Different tmp filesystem behavior\n\nRecommended fix:\n\n## Explicitly set execution environment\n\n```bash\n# First generation (legacy)\ngcloud run deploy my-service \\\n  --execution-environment=gen1\n\n# Second generation (recommended for most)\ngcloud run deploy my-service \\\n  --execution-environment=gen2\n```\n\n## Handle network differences\n\n```python\n# Second-gen doesn't auto-redirect HTTP to HTTPS\nfrom fastapi import FastAPI, Request\nfrom fastapi.responses import RedirectResponse\n\napp = FastAPI()\n\n@app.middleware(\"http\")\nasync def redirect_https(request: Request, call_next):\n    # Check X-Forwarded-Proto header\n    if request.headers.get(\"X-Forwarded-Proto\") == \"http\":\n        url = request.url.replace(scheme=\"https\")\n        return RedirectResponse(url, status_code=301)\n    return await call_next(request)\n```\n\n## GPU access (second-gen only)\n\n```bash\n# GPUs only available in second-gen\ngcloud run deploy ml-service \\\n  --execution-environment=gen2 \\\n  --gpu=1 \\\n  --gpu-type=nvidia-l4\n```\n\n## Check execution environment\n\n```python\nimport os\n\ndef get_execution_environment():\n    # Second-gen has different \u002Fproc structure\n    try:\n        with open('\u002Fproc\u002Fversion', 'r') as f:\n            version = f.read()\n            if 'gVisor' in version:\n                return 'gen2'\n    except:\n        pass\n    return 'gen1'\n```\n\n### Request Timeout Configuration Mismatch\n\nSeverity: MEDIUM\n\nSituation: Long-running requests or background processing\n\nSymptoms:\nRequests terminated before completion.\n504 Gateway Timeout errors.\nProcessing stops unexpectedly.\nInconsistent timeout behavior.\n\nWhy this breaks:\nCloud Run has multiple timeout configurations that must align:\n- Request timeout (default 300s, max 3600s for HTTP, 60m for gRPC)\n- Client timeout\n- Downstream service timeouts\n- Load balancer timeout (for external access)\n\nRecommended fix:\n\n## Set consistent timeouts\n\n```bash\n# Increase request timeout (max 3600s for HTTP)\ngcloud run deploy my-service \\\n  --timeout=900  # 15 minutes\n```\n\n## Handle long-running with webhooks\n\n```python\nfrom fastapi import FastAPI, BackgroundTasks\nimport httpx\n\napp = FastAPI()\n\n@app.post(\"\u002Fprocess\")\nasync def process(data: dict, background_tasks: BackgroundTasks):\n    task_id = create_task_id()\n\n    # Start background processing\n    background_tasks.add_task(\n        long_running_process,\n        task_id,\n        data,\n        data.get(\"callback_url\")\n    )\n\n    # Return immediately\n    return {\"task_id\": task_id, \"status\": \"processing\"}\n\nasync def long_running_process(task_id, data, callback_url):\n    result = await heavy_computation(data)\n\n    # Callback when done\n    if callback_url:\n        async with httpx.AsyncClient() as client:\n            await client.post(callback_url, json={\n                \"task_id\": task_id,\n                \"result\": result\n            })\n```\n\n## Use Cloud Tasks for reliable long-running\n\n```python\nfrom google.cloud import tasks_v2\n\ndef create_long_running_task(data):\n    client = tasks_v2.CloudTasksClient()\n    parent = client.queue_path(PROJECT, REGION, \"long-tasks\")\n\n    task = {\n        \"http_request\": {\n            \"http_method\": tasks_v2.HttpMethod.POST,\n            \"url\": \"https:\u002F\u002Fworker.run.app\u002Fprocess\",\n            \"body\": json.dumps(data).encode(),\n            \"headers\": {\"Content-Type\": \"application\u002Fjson\"}\n        },\n        \"dispatch_deadline\": {\"seconds\": 1800}  # 30 min\n    }\n\n    return client.create_task(parent=parent, task=task)\n```\n\n## Streaming for long responses\n\n```python\nfrom fastapi import FastAPI\nfrom fastapi.responses import StreamingResponse\n\n@app.get(\"\u002Flarge-report\")\nasync def large_report():\n    async def generate():\n        for chunk in process_large_data():\n            yield chunk\n\n    return StreamingResponse(generate(), media_type=\"text\u002Fplain\")\n```\n\n## Validation Checks\n\n### Hardcoded GCP Credentials\n\nSeverity: ERROR\n\nGCP credentials must never be hardcoded in source code\n\nMessage: Hardcoded GCP service account credentials. Use Secret Manager or Workload Identity.\n\n### GCP API Key in Source Code\n\nSeverity: ERROR\n\nAPI keys should use Secret Manager\n\nMessage: Hardcoded GCP API key. Use Secret Manager.\n\n### Credentials JSON File in Repository\n\nSeverity: ERROR\n\nService account JSON files should not be in source control\n\nMessage: Credentials file detected. Add to .gitignore and use Secret Manager.\n\n### Running as Root User\n\nSeverity: WARNING\n\nContainers should not run as root for security\n\nMessage: Dockerfile runs as root. Add USER directive for security.\n\n### Missing Health Check in Dockerfile\n\nSeverity: INFO\n\nCloud Run uses HTTP health checks, Dockerfile HEALTHCHECK is optional\n\nMessage: No HEALTHCHECK in Dockerfile. Cloud Run uses its own health checks.\n\n### Hardcoded Port in Application\n\nSeverity: WARNING\n\nPort should come from PORT environment variable\n\nMessage: Hardcoded port. Use PORT environment variable for Cloud Run.\n\n### Large File Writes to \u002Ftmp\n\nSeverity: WARNING\n\n\u002Ftmp uses container memory, large writes can cause OOM\n\nMessage: \u002Ftmp writes consume memory. Consider Cloud Storage for large files.\n\n### Synchronous File Operations\n\nSeverity: WARNING\n\nSync file ops block the event loop in async apps\n\nMessage: Synchronous file operations. Use async versions for better concurrency.\n\n### Global Mutable State\n\nSeverity: WARNING\n\nGlobal state issues with concurrent requests\n\nMessage: Global mutable state may cause issues with concurrent requests.\n\n### Thread-Unsafe Singleton Pattern\n\nSeverity: WARNING\n\nSingletons need thread safety for concurrency > 1\n\nMessage: Singleton pattern - ensure thread safety if using concurrency > 1.\n\n## Collaboration\n\n### Delegation Triggers\n\n- user needs AWS serverless -> aws-serverless (Lambda, API Gateway, SAM)\n- user needs Azure containers -> azure-functions (Azure Container Apps, Functions)\n- user needs database design -> postgres-wizard (Cloud SQL design, AlloyDB)\n- user needs authentication -> auth-specialist (Firebase Auth, Identity Platform)\n- user needs AI integration -> llm-architect (Vertex AI, Cloud Run + LLM)\n- user needs workflow orchestration -> workflow-automation (Cloud Workflows, Eventarc)\n\n## When to Use\nUse this skill when the request clearly matches the capabilities and patterns described above.\n\n## Limitations\n- Use this skill only when the task clearly matches the scope described above.\n- Do not treat the output as a substitute for environment-specific validation, testing, or expert review.\n- Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.\n",{"name":5,"risk":24,"source":25,"date_added":26,"description":7},"unknown","vibeship-spawner-skills (Apache 2.0)",1772150400000,"main","87afe6ce96220accf50a345dea46784c8c9f210e","YYC3-Skills-技能包\u002FYYC3-Skills-技能包-antigravity-awesome-skills\u002Fplugins\u002Fantigravity-awesome-skills-claude\u002Fskills\u002Fgcp-cloud-run",0,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base","https:\u002F\u002Fkb.yyc3.top\u002F","https:\u002F\u002Fgithub.com\u002FYYC-Cube","万象归元于云枢 | 深栈智启新纪元","2026-11-10 10:37:32",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill gcp-cloud-run",{"tool":42,"command":43},"bunx","bunx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill gcp-cloud-run",{"tool":45,"command":46},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill gcp-cloud-run","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-antigravity-awesome-skills\u002Fplugins\u002Fantigravity-awesome-skills-claude\u002Fskills\u002Fgcp-cloud-run\u002FSKILL.md",true,[],[51,58,65,72,79,86],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":19},"01M7CTDWHERCXPQCP0K03A86AS","vercel-sandbox","Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include \"Vercel Sandbox browser\", \"microVM Chrome\", \"agent-browser in sandbox\", \"browser automation on Vercel\", or any task requiring Chrome in a Vercel Sandbox.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fvercel-sandbox\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7CTD8D2BNJY54S65M47HEJA","slack","Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include \"check my Slack\", \"what channels have unreads\", \"send a message to\", \"search Slack for\", \"extract from Slack\", \"find who said\", or any task requiring programmatic Slack interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fslack\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7CTCSSB8ZP3Q58C3YZDZK1W","electron","Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include \"automate Slack app\", \"control VS Code\", \"interact with Discord app\", \"test this Electron app\", \"connect to desktop app\", or any task requiring automation of a native Electron application.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Felectron\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7CTCBJY76688V4DCZGWFHGQ","dogfood","Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to \"dogfood\", \"QA\", \"exploratory test\", \"find issues\", \"bug hunt\", \"test this app\u002Fsite\u002Fplatform\", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fdogfood\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7CTC4H0FFGVEAQR9PA4GA1A","agent-browser","Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to \"open a website\", \"fill out a form\", \"click a button\", \"take a screenshot\", \"scrape data from a page\", \"test this web app\", \"login to a site\", \"automate browser actions\", or any task requiring programmatic web interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fagent-browser\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7CTBKVFYBG78FDSABSDJTTQ","shadcn","Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn\u002Fui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for \"shadcn init\", \"create an app with --preset\", \"switch to --preset\", and for building a design system or applying a DESIGN.md or brand spec to shadcn\u002Fui.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-UI-界面组件\u002FYYC3-UI-界面组件-ui\u002Fskills\u002Fshadcn\u002FSKILL.md",[],[]]