N8N_AI_CloudRun Module — Configuration Guide

n8n is an open-source workflow automation platform that lets you connect services, run logic, and build AI-powered pipelines through a visual node-based interface. This module deploys n8n on Google Cloud Run alongside two companion AI services: Qdrant (vector database for RAG and document search) and Ollama (local LLM inference for privacy-first AI). Together they form an AI Starter Kit for building intelligent agents, chatbots, and document analysis workflows without external AI API dependencies.

N8N_AI_CloudRun is a wrapper module built on top of App_CloudRun. It uses App_CloudRun for all GCP infrastructure provisioning (Cloud Run service, networking, Cloud SQL, GCS, secrets, CI/CD) and adds n8n-specific application configuration and AI component orchestration on top.

Note: Variables marked as platform-managed are set and maintained by the platform. You do not normally need to change them.

---

How This Guide Is Structured

This guide documents only the variables that are unique to N8N_AI_CloudRun or that have n8n-specific defaults that differ from the App_CloudRun base module. For all other variables — project identity, CI/CD, GCS storage, backup, custom SQL, access and networking, IAP, Cloud Armor, and VPC Service Controls — refer directly to the App_CloudRun Configuration Guide.

Variables fully covered by the App_CloudRun guide:

Configuration Area	App_CloudRun_Guide Section	N8N_AI-Specific Notes
Module Metadata & Configuration	Group 0	Different defaults for module_description and module_documentation.
Project & Identity	Group 1	Refer to base App_CloudRun module documentation.
Application Identity	Group 2	See N8N AI Application Identity below for n8n-specific defaults.
Runtime & Scaling	Group 3	See N8N Runtime Configuration below. container_port defaults to 5678. cpu_limit and memory_limit are top-level variables.
Environment Variables & Secrets	Group 4	See N8N Environment Variables below for SMTP defaults.
Observability & Health	Group 5	See N8N Health Probes below for n8n-specific probe defaults.
Jobs & Scheduled Tasks	Group 6	Refer to base App_CloudRun module documentation.
CI/CD & GitHub Integration	Group 7	Refer to base App_CloudRun module documentation.
Storage — NFS	Group 8	NFS is enabled by default (enable_nfs = true). See Platform-Managed Behaviours.
Storage — GCS	Group 9	Refer to base App_CloudRun module documentation.
Redis Cache	Group 10	See Redis Configuration below — n8n adds enable_redis and redis_host toggles not present in the base module.
Database Backend	Group 11	See N8N Database Configuration below. db_name and db_user replace application_database_name and application_database_user.
Backup & Maintenance	Group 12	Refer to base App_CloudRun module documentation.
Custom Initialisation & SQL	Group 13	Refer to base App_CloudRun module documentation.
Access & Networking	Group 14	Refer to base App_CloudRun module documentation.
Identity-Aware Proxy	Group 15	Refer to base App_CloudRun module documentation. Note: enabling IAP blocks public webhook endpoints.
Cloud Armor & CDN	Group 16	Refer to base App_CloudRun module documentation.
VPC Service Controls	Group 17	Refer to base App_CloudRun module documentation.

---

Platform-Managed Behaviours

The following behaviours are applied automatically by N8N_AI_CloudRun regardless of the variable values in your tfvars file. They cannot be overridden by user configuration.

Behaviour	Detail
Encryption key auto-generated	A 32-character random encryption key is generated and stored in Secret Manager as N8N_ENCRYPTION_KEY. This key encrypts all n8n credentials (API keys, OAuth tokens, passwords stored in workflows). It is injected into the Cloud Run service automatically. Back up this secret before destroying the module — credentials encrypted with one key cannot be decrypted with a different key.
SMTP password auto-generated	A placeholder SMTP password is generated and stored in Secret Manager as N8N_SMTP_PASS. Replace the secret value with your real SMTP credentials before enabling email sending.
n8n port fixed at 5678	N8N_PORT=5678 is injected automatically. The container_port variable defaults to 5678 to match. Do not override N8N_PORT in environment_variables.
Database type set to PostgreSQL	DB_TYPE=postgresdb is injected automatically. n8n requires PostgreSQL — do not change database_type to MySQL or SQL Server.
Database connection variables injected	DB_POSTGRESDB_HOST, DB_POSTGRESDB_PORT, DB_POSTGRESDB_DATABASE, DB_POSTGRESDB_USER, and DB_POSTGRESDB_PASSWORD are injected automatically from the Cloud SQL instance provisioned by App_CloudRun.
Webhook and editor URLs auto-set	WEBHOOK_URL and N8N_EDITOR_BASE_URL are set to the predicted Cloud Run service URL, computed from the project number and deployment region before the service is created. This allows n8n webhooks to be correctly advertised in the UI.
Qdrant URL auto-injected	When enable_qdrant = true, QDRANT_URL is set to the internal URL of the companion Qdrant Cloud Run service. This URL is accessible only within the VPC — Qdrant is not exposed to the public internet.
Ollama host auto-injected	When enable_ollama = true, OLLAMA_HOST is set to the internal URL of the companion Ollama Cloud Run service. Ollama is not exposed to the public internet.
Qdrant and Ollama use internal-only ingress	The Qdrant and Ollama Cloud Run services are deployed with INGRESS_TRAFFIC_INTERNAL_ONLY. They can only be reached from within the VPC by the n8n service.
GCS persistence for AI data	Qdrant stores its vector index at /mnt/gcs/qdrant and Ollama stores model weights at /mnt/gcs/ollama/models via GCS Fuse on the shared -n8n-data bucket. This persists data across container restarts.
Database initialisation job	A Cloud Run Job (db-init) is created automatically to provision the n8n_db database and n8n_user PostgreSQL user before the n8n container starts.

---

N8N AI Application Identity

These variables control how the n8n deployment is named and described. They correspond to Group 2 variables in App_CloudRun but carry n8n-specific defaults.

Variable	Default	Options / Format	Description & Implications
application_name	"n8nai"	[a-z][a-z0-9-]{0,19}	Internal identifier used as the base name for the Cloud Run service, Artifact Registry repository, Secret Manager secrets, and GCS buckets. Do not change after initial deployment — it is embedded in resource names and changing it will cause resources to be recreated.
application_display_name	"N8N AI Starter Kit"	Any string	Human-readable name shown in the platform UI, the Cloud Run service list, and monitoring dashboards. Can be updated freely without affecting resource names.
description	"N8N AI Starter Kit - Workflow automation with Qdrant and Ollama"	Any string	Brief description of the deployment. Populated into the Cloud Run service description field and platform documentation.
application_version	"2.4.7"	n8n version string, e.g. "2.4.7", "latest"	Version tag applied to the container image and used for deployment tracking. Increment this value to trigger a new image build and revision. See n8n releases for available versions.

Validating Application Identity

# Confirm the Cloud Run service exists with the expected name
gcloud run services describe n8nai \
  --region=REGION \
  --format="table(metadata.name,metadata.annotations['run.googleapis.com/description'])"

---

N8N Runtime Configuration

n8n listens on port 5678 and exposes cpu_limit and memory_limit as dedicated top-level variables rather than requiring users to set the full container_resources object.

Variable	Default	Options / Format	Description & Implications
container_port	5678	Integer, 1–65535	The TCP port n8n binds to inside the container. Cloud Run routes incoming traffic to this port. Must match N8N_PORT. Do not change unless you are overriding the default n8n port.
cpu_limit	"2000m"	Cloud Run CPU string (e.g. "1000m", "2000m", "4")	CPU limit for the n8n container. 2 vCPU is recommended for active AI workflows. n8n executes workflow nodes concurrently and the AI node operations (vector search, LLM calls) are CPU-bound. Setting below "1000m" risks throttling on complex workflows.
memory_limit	"4Gi"	Cloud Run memory string (e.g. "2Gi", "4Gi", "8Gi")	Memory limit for the n8n container. 4 Gi is recommended for AI workflows. n8n caches workflow state and credential data in memory; AI nodes processing large document sets can require 2–3 Gi alone.
min_instance_count	0	Integer ≥ 0	Minimum running instances. Set to 0 to enable scale-to-zero when idle (lowest cost). Set to 1 to eliminate cold starts and ensure webhook availability — n8n webhooks registered in the platform are only active while at least one instance is running.
max_instance_count	1	Integer ≥ 1	Maximum concurrent instances. The default of 1 ensures workflow state consistency. Increase only after configuring Redis queue mode (enable_redis = true) — without Redis, multiple instances will conflict on credential and workflow state.
timeout_seconds	300	Integer, 0–3600	Maximum request duration before Cloud Run returns a 504 timeout. n8n long-running workflow executions or Ollama inference requests can exceed the default. Increase to 600 or 900 for workflows that call large LLMs or process many documents.

Note on container_resources: The full container_resources object (as documented in App_CloudRun_Guide Group 3) is also available. If container_resources is set explicitly in your tfvars, it takes precedence over the top-level cpu_limit and memory_limit variables. Use container_resources when you also need to set cpu_request or mem_request.

N8N-specific runtime defaults that differ from App_CloudRun:

Variable	App_CloudRun Default	N8N_AI_CloudRun Default	Reason
container_port	8080	5678	n8n's native port.
cpu_limit	"1000m"	"2000m"	AI workflows are CPU-intensive.
memory_limit	"512Mi"	"4Gi"	n8n with AI nodes requires substantial memory.
max_instance_count	1	1	Multi-instance requires Redis queue mode.
enable_nfs	false	true	NFS provides shared persistence for workflow data and credentials across restarts.
enable_cloudsql_volume	true	true	n8n connects to Cloud SQL via the Auth Proxy Unix socket.

Validating Runtime Configuration

# View the CPU, memory, and port on the latest revision
gcloud run services describe n8nai \
  --region=REGION \
  --format="yaml(spec.template.spec.containers[0].resources,spec.template.spec.containers[0].ports)"

---

N8N Health Probes

The N8N_AI_CloudRun module uses flat probe objects (startup_probe and liveness_probe) that differ from the startup_probe_config / health_check_config structured objects documented in App_CloudRun_Guide Group 5.

Variable	Default	Description & Implications
startup_probe	{ enabled=true, type="HTTP", path="/", initial_delay_seconds=120, timeout_seconds=3, period_seconds=10, failure_threshold=3 }	Determines when n8n is ready to receive traffic after starting. The initial_delay_seconds=120 gives n8n time to connect to Cloud SQL and load workflow state before the probe begins. Reduce this if your n8n instance starts quickly (small workflow count, warm database connection).
liveness_probe	{ enabled=true, type="HTTP", path="/", initial_delay_seconds=30, timeout_seconds=5, period_seconds=30, failure_threshold=3 }	Periodically checks that the running n8n container is healthy. A failed liveness probe causes Cloud Run to restart the container. The initial_delay_seconds=30 avoids false restarts during startup.

Note: The startup_probe_config and health_check_config structured object variables are also accepted by this module. When both the flat and structured forms are provided, the structured form takes precedence.

---

AI Components Configuration

These variables are unique to N8N_AI_CloudRun — they do not exist in App_CloudRun. They control the Qdrant vector database and Ollama LLM server that are deployed as companion Cloud Run services.

Variable	Default	Options / Format	Description & Implications
enable_ai_components	true	true / false	Master toggle for the entire AI stack. Set to false to deploy n8n as a standard workflow automation tool without Qdrant or Ollama. When false, the QDRANT_URL and OLLAMA_HOST environment variables are not injected. Individual components can be further controlled via enable_qdrant and enable_ollama.
enable_qdrant	true	true / false	Deploys Qdrant vector database as a companion Cloud Run service for n8n AI workflows. Qdrant enables RAG pipelines, document embedding search, and AI memory. Only used when enable_ai_components = true. When false, the QDRANT_URL variable is not injected and the Qdrant service is not created.
qdrant_version	"latest"	Docker image tag, e.g. "latest", "v1.9.0"	Image version of the qdrant/qdrant container. Use a pinned version (e.g. "v1.9.0") in production for reproducible deployments. Only used when enable_qdrant = true.
enable_ollama	true	true / false	Deploys Ollama LLM server as a companion Cloud Run service. Ollama runs open-source models (Llama 3, Mistral, Gemma) on your infrastructure — no external AI API keys required. Only used when enable_ai_components = true. Ollama requires at least 4 Gi of memory on the companion service.
ollama_version	"latest"	Docker image tag, e.g. "latest", "0.3.0"	Image version of the ollama/ollama container. Use a pinned version in production. Only used when enable_ollama = true.
ollama_model	"llama3.2"	Ollama model name, e.g. "llama3.2", "mistral", "gemma2"	The default language model served by Ollama. This model is available to n8n AI nodes for text generation, summarisation, and chat workflows. Larger models (e.g. "llama3:70b") require more CPU and memory than the Ollama service defaults.

AI Component Resource Allocation

The Qdrant and Ollama services are deployed with fixed resources managed by the platform. These are not user-configurable in this release:

Service	CPU	Memory	Scaling	Storage
Qdrant	1 vCPU	1 Gi	Fixed: 1 instance	GCS Fuse at /mnt/gcs/qdrant
Ollama	2 vCPU	4 Gi	Fixed: 1 instance	GCS Fuse at /mnt/gcs/ollama/models

Note on GPU support: Ollama currently runs on CPU only. When Cloud Run GPU support becomes generally available, a future release will add GPU acceleration for faster LLM inference.

Validating AI Components

# List all Cloud Run services in the project (n8n, qdrant, and ollama should appear)
gcloud run services list --project=PROJECT_ID --region=REGION \
  --format="table(name,status.url,status.conditions[0].type)"

# Confirm QDRANT_URL and OLLAMA_HOST are injected into the n8n service
gcloud run services describe n8nai \
  --region=REGION \
  --format="yaml(spec.template.spec.containers[0].env)" | grep -E "QDRANT|OLLAMA"

# Check the Qdrant service is internal-only
gcloud run services describe QDRANT_SERVICE_NAME \
  --region=REGION \
  --format="yaml(metadata.annotations['run.googleapis.com/ingress'])"

---

Redis Configuration

These variables are unique to N8N_AI_CloudRun at the module level. The base App_CloudRun module accepts redis_auth but does not have the enable_redis, redis_host, or redis_port toggles. Redis is required for n8n queue mode, which enables reliable multi-instance workflow execution.

Variable	Default	Options / Format	Description & Implications
enable_redis	true	true / false	Enables Redis as the n8n queue mode backend by injecting REDIS_HOST and REDIS_PORT into the Cloud Run service. When true and redis_host is empty, the module defaults to the NFS server IP (if one is discovered via the Services_GCP NFS discovery). Required when max_instance_count > 1 to avoid workflow state conflicts between instances.
redis_host	"" (auto-discovered)	Hostname or IP, e.g. "10.0.0.5", "redis.internal"	Hostname or IP of the Redis server. Leave blank to use the NFS server IP auto-discovered from Services_GCP. Override with a dedicated Redis/Memorystore instance endpoint for production deployments requiring higher availability or AUTH.
redis_port	"6379"	Port string, e.g. "6379"	TCP port of the Redis server. Must match the port configured on the Redis instance.
redis_auth	""	Sensitive string	Authentication password for the Redis server. Leave empty for unauthenticated Redis. For Google Cloud Memorystore with AUTH enabled, set this to the instance auth string. Treated as sensitive — never stored in Terraform state in plaintext.

For full documentation of the Redis Cache group including Memorystore provisioning and TLS configuration, refer to App_CloudRun_Guide Group 10.

Validating Redis Configuration

# Confirm REDIS_HOST and REDIS_PORT are injected into the n8n service
gcloud run services describe n8nai \
  --region=REGION \
  --format="yaml(spec.template.spec.containers[0].env)" | grep -E "REDIS"

---

N8N Database Configuration

n8n requires PostgreSQL. This module exposes db_name and db_user as short top-level variables in place of the application_database_name and application_database_user variables documented in App_CloudRun_Guide Group 11.

All other database variables (database_password_length, enable_auto_password_rotation, rotation_propagation_delay_sec, secret_rotation_period, etc.) behave identically to the App_CloudRun equivalents — refer to App_CloudRun_Guide Group 11 for their documentation.

Variable	Default	Options / Format	Description & Implications
db_name	"n8n_db"	[a-z][a-z0-9_]{0,62}	Name of the PostgreSQL database created within the Cloud SQL instance. Injected automatically as DB_POSTGRESDB_DATABASE. Do not change after initial deployment — renaming the database requires a full backup-and-restore migration.
db_user	"n8n_user"	[a-z][a-z0-9_]{0,31}	PostgreSQL user created for n8n. Injected automatically as DB_POSTGRESDB_USER. The password is auto-generated, stored in Secret Manager, and injected as DB_POSTGRESDB_PASSWORD.

Validating Database Configuration

# Confirm the database and user were created
gcloud sql databases list --instance=INSTANCE_NAME --project=PROJECT_ID

gcloud sql users list --instance=INSTANCE_NAME --project=PROJECT_ID

# Confirm database env vars are injected into the Cloud Run service
gcloud run services describe n8nai \
  --region=REGION \
  --format="yaml(spec.template.spec.containers[0].env)" | grep -E "DB_POSTGRES"

---

N8N Environment Variables

The environment_variables variable (documented in App_CloudRun_Guide Group 4) has n8n-specific defaults that configure email delivery. These are plain-text values — for the SMTP password use secret_environment_variables.

Default environment_variables in N8N_AI_CloudRun:

environment_variables = {
  SMTP_HOST     = ""
  SMTP_PORT     = "25"
  SMTP_USER     = ""
  SMTP_PASSWORD = ""
  SMTP_SSL      = "false"
  EMAIL_FROM    = "ghost@example.com"
}

Override the SMTP values to enable n8n email notifications (workflow failure alerts, credential sharing invitations). For providers such as SendGrid or Mailgun that use API key authentication, set SMTP_USER = "apikey" and store the actual key in secret_environment_variables.

Do not set N8N_PORT, DB_TYPE, DB_POSTGRESDB_*, N8N_ENCRYPTION_KEY, WEBHOOK_URL, N8N_EDITOR_BASE_URL, QDRANT_URL, or OLLAMA_HOST in environment_variables — these are injected automatically by the platform and will be overridden.

---

Configuration Examples

Basic Deployment

Deploys n8n with AI components using default settings. Suitable for evaluation and development.

# config/basic.tfvars
resource_creator_identity = ""
project_id                = "my-project-123"
tenant_deployment_id      = "basic"

Advanced Deployment

Production-grade deployment with scaled resources, Redis queue mode, CI/CD, and full observability.

# config/advanced.tfvars
resource_creator_identity = ""
project_id                = "my-project-123"
tenant_deployment_id      = "prod"

application_name         = "n8nai"
application_display_name = "N8N AI Production"

# Scaling & Performance
cpu_limit          = "4000m"
memory_limit       = "8Gi"
min_instance_count = 1
max_instance_count = 5

# AI Components
enable_ai_components = true
enable_qdrant        = true
qdrant_version       = "v1.9.0"
enable_ollama        = true
ollama_version       = "0.3.0"
ollama_model         = "llama3.2"

# Redis (required for multi-instance scaling)
enable_redis = true

# Database
database_password_length = 32

# CI/CD & Cloud Deploy
enable_cicd_trigger = true
enable_cloud_deploy = true
cloud_deploy_stages = [
  { name = "dev",     require_approval = false, auto_promote = false },
  { name = "staging", require_approval = false, auto_promote = false },
  { name = "prod",    require_approval = true,  auto_promote = false },
]

# Security
enable_iap                  = false  # Note: enabling IAP blocks public webhooks
enable_binary_authorization = true
enable_cloud_armor          = true

# Backup
backup_schedule       = "0 2 * * *"
backup_retention_days = 30

# Observability
uptime_check_config = {
  enabled        = true
  path           = "/"
  check_interval = "60s"
  timeout        = "10s"
}

Custom Image Deployment

Deploys n8n with a custom-built container image, external SMTP, and Redis configured explicitly.

# config/custom.tfvars
resource_creator_identity = ""
project_id                = "my-project-123"
tenant_deployment_id      = "custom"

application_name = "n8nai"

# Custom Container Build
container_image_source = "custom"
container_build_config = {
  enabled            = true
  dockerfile_path    = "Dockerfile"
  context_path       = "scripts"
  dockerfile_content = null
  build_args         = {}
  artifact_repo_name = "n8n-repo"
}

# AI Components
enable_ai_components = true
enable_qdrant        = true
enable_ollama        = true

# Redis
enable_redis = true
redis_host   = "10.0.0.5"   # Explicit Memorystore IP

# SMTP
environment_variables = {
  SMTP_HOST  = "smtp.sendgrid.net"
  SMTP_PORT  = "587"
  SMTP_USER  = "apikey"
  SMTP_SSL   = "true"
  EMAIL_FROM = "noreply@example.com"
}

secret_environment_variables = {
  SMTP_PASSWORD = "sendgrid-api-key-secret"
}