N8N AI on GKE Autopilot

n8n is an open-source workflow automation platform with a visual node-based interface for connecting services, running logic, and building AI-powered pipelines. This module deploys n8n on GKE Autopilot alongside two companion AI services — Qdrant (vector database for RAG and semantic search) and Ollama (local LLM inference for privacy-first AI) — on top of the App_GKE foundation, which provisions and manages the shared Google Cloud and Kubernetes infrastructure.

This guide focuses on the cloud services n8n AI uses and how to explore and operate them from the Google Cloud Console and the command line. For the mechanics common to every GKE application — Workload Identity, ingress, autoscaling, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC Service Controls, backups, and the deployment lifecycle — refer to the App_GKE foundation guide rather than repeating them here.

---

1. Overview

N8N AI runs as a Node.js workflow workload alongside Qdrant and Ollama Kubernetes Deployments. The deployment wires together a focused set of Google Cloud services:

Capability	Google Cloud service	Notes
Compute	GKE Autopilot	n8n + Qdrant + Ollama Deployments, 2 vCPU / 4 GiB by default for n8n
Database	Cloud SQL for PostgreSQL 15	Required — n8n requires PostgreSQL; the engine is fixed
Object storage	Cloud Storage (GCS Fuse)	Shared AI data bucket mounted at /mnt/gcs; workflow data at /home/node/.n8n
Shared files	Filestore (NFS)	Shared NFS volume for cross-replica persistence; doubles as default Redis host
Cache & queue	Redis	Enabled by default; used for n8n queue mode across multiple replicas
Vector database	Qdrant (in-cluster)	Deployed as a companion Kubernetes Deployment; internal-only ClusterIP service
LLM inference	Ollama (in-cluster)	Deployed as a companion Kubernetes Deployment; internal-only ClusterIP service
Secrets	Secret Manager	Auto-generated N8N_ENCRYPTION_KEY and N8N_SMTP_PASS; synced to Kubernetes Secrets
Ingress	Cloud Load Balancing	External LoadBalancer, optional custom domain + managed certificate via Gateway API

Sensible defaults worth knowing up front:

• PostgreSQL 15 is mandatory. The database engine is fixed by the common configuration; changing database_type to MySQL breaks startup.
• N8N_ENCRYPTION_KEY is auto-generated and stored in Secret Manager. Back it up before destroying the module — credentials encrypted with one key cannot be decrypted with a different key.
• Qdrant and Ollama run as internal-only Kubernetes Deployments in the same namespace as n8n, each with a ClusterIP service. They are not exposed outside the cluster.
• GCS Fuse persistence keeps Qdrant's vector index (/mnt/gcs/qdrant) and Ollama's model weights (/mnt/gcs/ollama/models) durable across pod restarts.
• Redis is enabled by default. With more than one replica, a shared queue backend is required to prevent split-brain workflow execution.
• min_instance_count defaults to 0. GKE HPA does not support true scale-to-zero the same way Cloud Run does; set to 1 for reliable webhook availability.

---

2. Google Cloud Services & How to Explore Them

All commands assume you have run gcloud container clusters get-credentials <cluster> --region <region> --project <project> and that PROJECT, REGION, and NAMESPACE are set. The namespace and other identifiers are reported in the deployment Outputs.

A. GKE Autopilot — the n8n AI workload

n8n, Qdrant, and Ollama each run as a separate Kubernetes Deployment in the same namespace. Horizontal Pod Autoscaling governs n8n replica count between the minimum and maximum limits. Qdrant and Ollama are fixed at one replica each.

• Console: Kubernetes Engine → Workloads → select each workload to see pods, revisions, and events. Kubernetes Engine → Services & Ingress shows external IPs and ClusterIP endpoints.
• CLI:

  kubectl get deployments,svc,hpa -n "$NAMESPACE"
  kubectl logs -n "$NAMESPACE" deploy/n8nai --tail=100
  kubectl logs -n "$NAMESPACE" deploy/qdrant --tail=50
  kubectl logs -n "$NAMESPACE" deploy/ollama --tail=50
  kubectl describe hpa -n "$NAMESPACE"

See App_GKE for Autopilot, scaling, and workload type details.

B. Cloud SQL for PostgreSQL 15

n8n stores all workflow definitions, execution history, and credentials in a managed Cloud SQL for PostgreSQL 15 instance. Pods reach it privately through the Cloud SQL Auth Proxy sidecar over a Unix socket at 127.0.0.1. On first deploy a db-init job creates the application database and user.

• Console: SQL → select the instance for connections, backups, flags, and metrics.
• CLI:

  gcloud sql instances list --project "$PROJECT"
  gcloud sql instances describe <instance-name> --project "$PROJECT"
  gcloud sql connect <instance-name> --user=<db-user> --database=n8n_db --project "$PROJECT"

The instance name, database name, user, and the Secret Manager secret holding the password are all surfaced in the Outputs. For the connection model, automated backups, and password rotation, see App_GKE.

C. Cloud Storage (GCS Fuse) and Filestore (NFS)

A shared Cloud Storage bucket is mounted into every container via GCS Fuse:

• n8n workflow data and credentials at /home/node/.n8n
• Qdrant's vector index at /mnt/gcs/qdrant
• Ollama's model weights at /mnt/gcs/ollama/models

Filestore (NFS) is enabled by default and provides a shared persistent volume for multi-replica data and doubles as the default Redis endpoint.

• Console: Cloud Storage → Buckets for the AI data bucket; Filestore → Instances for the NFS share.
• CLI:

  gcloud storage buckets list --project "$PROJECT"
  gcloud storage ls gs://<ai-data-bucket>/          # bucket name is in the Outputs
  gcloud filestore instances list --project "$PROJECT"
  kubectl exec -n "$NAMESPACE" deploy/n8nai -- df -h | grep -E "gcs|nfs"

See App_GKE for NFS provisioning, GCS Fuse, and CMEK options.

D. Redis queue backend

Redis backs n8n's queue mode, enabling reliable multi-replica workflow execution. When no redis_host is configured and NFS is enabled, the NFS server IP is used as the Redis endpoint. For production with high availability, point to a Cloud Memorystore instance.

• Console: Memorystore → Redis (if using a managed instance).
• CLI:

  kubectl describe pod -n "$NAMESPACE" -l app=n8nai | grep -E "REDIS"
  redis-cli -h <redis-host> ping

E. Qdrant — vector database

Qdrant provides high-performance vector similarity search for RAG pipelines, document embeddings, and AI memory within n8n workflows. It is deployed as an internal-only Kubernetes Deployment and is reachable from n8n via the QDRANT_URL environment variable, which points to its ClusterIP service.

• Console: Kubernetes Engine → Workloads → select the qdrant Deployment.
• CLI:

  kubectl get deployment qdrant -n "$NAMESPACE"
  kubectl logs -n "$NAMESPACE" deploy/qdrant --tail=50
  # Confirm QDRANT_URL is injected into n8n
  kubectl describe pod -n "$NAMESPACE" -l app=n8nai | grep QDRANT_URL
  # Check Qdrant's health endpoint from inside the cluster
  kubectl exec -n "$NAMESPACE" deploy/n8nai -- curl -s http://qdrant:6333/healthz

F. Ollama — local LLM server

Ollama runs open-source language models (Llama 3, Mistral, Gemma) directly on your infrastructure, enabling privacy-first AI inference without external API dependencies. It is an internal-only Kubernetes Deployment; OLLAMA_HOST is injected into n8n to reach it.

• Console: Kubernetes Engine → Workloads → select the ollama Deployment.
• CLI:

  kubectl get deployment ollama -n "$NAMESPACE"
  kubectl logs -n "$NAMESPACE" deploy/ollama --tail=50
  # Confirm OLLAMA_HOST is injected into n8n
  kubectl describe pod -n "$NAMESPACE" -l app=n8nai | grep OLLAMA_HOST
  # List models loaded in Ollama (from inside the cluster)
  kubectl exec -n "$NAMESPACE" deploy/n8nai -- curl -s http://ollama:11434/api/tags

G. Secret Manager

The n8n encryption key and SMTP password are generated automatically and stored as Secret Manager secrets, then synced to Kubernetes Secrets and injected into pods at runtime.

• Console: Security → Secret Manager.
• CLI:

  gcloud secrets list --project "$PROJECT"
  gcloud secrets versions access latest --secret=<encryption-key-secret> --project "$PROJECT"

The database password secret name is in the Outputs. See App_GKE for the Secret Store CSI integration and rotation.

H. Networking & ingress

By default the workload is exposed through an external Cloud Load Balancing IP. Custom domain support via the Kubernetes Gateway API with a Google-managed certificate is enabled by default (enable_custom_domain = true), and a static IP is reserved by default so the address survives redeploys.

• Console: Network services → Load balancing; VPC network → IP addresses.
• CLI:

  kubectl get gateway,httproute,svc -n "$NAMESPACE"
  gcloud compute addresses list --project "$PROJECT"

See App_GKE for custom domains, Cloud CDN, and static IP.

I. Cloud Logging & Monitoring

Pod stdout/stderr flows to Cloud Logging; GKE and Cloud SQL metrics flow to Cloud Monitoring. Optional uptime checks and alert policies are available.

• Console: Logging → Logs Explorer; Monitoring → Dashboards / Alerting.
• CLI:

  gcloud logging read \
    'resource.type="k8s_container" AND resource.labels.namespace_name="'"$NAMESPACE"'"' \
    --project "$PROJECT" --limit 50

---

3. N8N AI Application Behaviour

• First-deploy database setup. A db-init Kubernetes Job runs before the n8n Deployment starts. It creates the n8n_db PostgreSQL database and n8n_user user via the Cloud SQL Auth Proxy socket, grants full privileges, and shuts down the proxy cleanly. The job is idempotent and safe to re-run.
• Encryption key. N8N_ENCRYPTION_KEY is auto-generated on first deploy and stored in Secret Manager. Back up this secret before destroying the module. All n8n credentials (API keys, OAuth tokens, workflow passwords) are encrypted with this key; they cannot be decrypted after a re-deploy with a different key.
• Health probes. The startup probe targets GET / on port 5678 with a 120-second initial delay, giving n8n time to connect to PostgreSQL and load workflow state. The liveness probe checks the same path after startup.
• Webhook and editor URLs. WEBHOOK_URL and N8N_EDITOR_BASE_URL are set to the predicted service URL before the Deployment is created so webhooks work without a post-deploy re-apply.
• Queue mode. When Redis is enabled, n8n operates in queue mode for reliable workflow execution across multiple replicas. Without Redis, only a single replica should run.
• SMTP password. N8N_SMTP_PASS is auto-generated as a placeholder. Replace the secret value in Secret Manager with real SMTP credentials before enabling email sending.
• Inspect scheduled jobs and CronJobs:

  kubectl get jobs -n "$NAMESPACE" --sort-by=.metadata.creationTimestamp
  kubectl get cronjobs -n "$NAMESPACE"

---

4. Configuration Variables

Variables are grouped exactly as they appear on the deployment platform. Only settings specific to or notable for n8n AI are listed; every other input is inherited from App_GKE with its standard behaviour and defaults.

Group 1 — Project & Identity

Variable	Default	Description
project_id	(required)	Target Google Cloud project.
region	us-central1	Region for the workload and regional resources.

Group 2 — Deployment Environment

Variable	Default	Description
tenant_deployment_id	demo	Short suffix that makes resource names unique per environment.
support_users	[]	Emails granted project access and monitoring alerts.
resource_labels	{}	Labels applied to all resources for cost/ownership tracking.

Group 3 — Application Identity

Variable	Default	Description
application_name	n8nai	Base name for resources. Do not change after first deploy.
application_display_name	N8N AI Starter Kit	Friendly name shown in the Console.
description	(set)	Workload description annotation.
application_version	2.4.7	n8n image version tag; increment to roll out a new version.

Group 4 — Runtime & Scaling

Variable	Default	Description
deploy_application	true	Set false to provision infrastructure only.
cpu_limit	2000m	CPU per n8n pod; also inherited by Ollama. 2 vCPU recommended.
memory_limit	4Gi	Memory per n8n pod; also inherited by Ollama. 4 GiB minimum for AI workloads.
min_instance_count	0	Minimum n8n replicas. Set to 1 for continuous webhook availability.
max_instance_count	3	Maximum replicas. Increase only with Redis enabled.
enable_cloudsql_volume	true	Cloud SQL Auth Proxy sidecar for socket connections.
enable_vertical_pod_autoscaling	false	Let Autopilot tune resource requests automatically.
timeout_seconds	300	Max request duration; increase for long AI inference workflows.

Group 5 — Environment Variables & Secrets

Variable	Default	Description
environment_variables	SMTP placeholders	Non-sensitive settings. Core n8n vars are injected automatically; do not set N8N_PORT, DB_TYPE, DB_POSTGRESDB_*, N8N_ENCRYPTION_KEY, WEBHOOK_URL, QDRANT_URL, or OLLAMA_HOST.
secret_environment_variables	{}	Map of env var → Secret Manager secret name. Use for external AI provider API keys.

Group 6 — GKE Backend & Cluster

Variable	Default	Description
service_type	LoadBalancer	How the Service is exposed.
session_affinity	None	n8n AI is stateless across instances with Redis. Change to ClientIP only if needed.
workload_type	null	Auto-resolves to StatefulSet when per-pod storage is enabled.
network_tags	['nfsserver']	Required for NFS connectivity firewall rules.
gke_cluster_name	""	Leave empty for auto-discovery.
namespace_name	""	Leave empty to auto-generate.

Group 7 — StatefulSet

Variable	Default	Description
stateful_pvc_enabled	null	Enable PVC templates in the StatefulSet. Setting true auto-selects StatefulSet.
stateful_pvc_size	10Gi	Storage size for each PVC.
stateful_pvc_mount_path	/data	Container path where the PVC is mounted.
stateful_pvc_storage_class	standard-rwo	Kubernetes StorageClass for PVCs.

Group 8 — Resource Quota

Variable	Default	Description
enable_resource_quota	false	Cap namespace CPU/memory/object counts.
quota_memory_requests / quota_memory_limits	""	Must use binary units (4Gi, 8192Mi) — bare integers are treated as bytes and block scheduling.

Group 9 — Reliability Policies

Variable	Default	Description
enable_pod_disruption_budget	true	Protect availability during node upgrades.
pdb_min_available	1	Raise min_instance_count above 1 if you need eviction headroom.
enable_topology_spread	false	Spread n8n, Qdrant, and Ollama pods across zones.

Group 10 — Observability & Health

Variable	Default	Description
startup_probe	HTTP / — 120s delay	n8n startup probe; generous delay allows DB connect + workflow load.
liveness_probe	HTTP / — 30s delay	n8n liveness probe.
startup_probe_config	TCP — enabled	App_GKE-standard startup probe.
health_check_config	HTTP / — enabled	App_GKE-standard liveness probe.
uptime_check_config	disabled	Enable for production monitoring.
alert_policies	[]	Optional metric alert policies.

Group 11 — Workload Automation

Variable	Default	Description
initialization_jobs	[]	Leave empty to use the built-in db-init job.
cron_jobs	[]	Scheduled CronJobs for periodic tasks (exports, maintenance).
additional_services	[]	Extra sidecar or helper Deployments alongside n8n.

Group 12 — CI/CD & GitHub Integration

Standard App_GKE Cloud Build / Cloud Deploy integration — see App_GKE. Key inputs: enable_cicd_trigger, github_repository_url, github_token, enable_cloud_deploy.

Group 13 — Filesystem (NFS)

Variable	Default	Description
enable_nfs	true	Shared Filestore volume for workflow data and default Redis host discovery.
nfs_mount_path	/mnt/nfs	Mount path inside the container.

Group 14 — Cloud Storage & Artifact Registry

Variable	Default	Description
create_cloud_storage	true	Provision GCS buckets.
storage_buckets	[]	Additional buckets beyond the auto-provisioned AI data bucket.
gcs_volumes	[]	Additional GCS Fuse mounts.
manage_storage_kms_iam / enable_artifact_registry_cmek	false	CMEK options.

Group 15 — Redis Queue Backend

Variable	Default	Description
enable_redis	true	Use Redis for n8n queue mode. Required when max_instance_count > 1.
redis_host	""	Leave empty to use the NFS server IP; set explicitly for a Memorystore instance.
redis_port	6379	Redis port.
redis_auth	""	Optional Redis auth password (sensitive).

Group 16 — Database Backend

Variable	Default	Description
db_name	n8n_db	PostgreSQL database name. Immutable after first deploy.
db_user	n8n_user	Application user. Immutable after first deploy.
database_password_length	32	Generated password length (16–64).
enable_auto_password_rotation	false	Zero-downtime DB password rotation.

Group 17 — Backup & Maintenance

Variable	Default	Description
backup_schedule	0 2 * * *	Automated backup cron (UTC).
backup_retention_days	7	Retention; raise to 30–90 for production/compliance.
enable_backup_import / backup_source / backup_uri	restore options	Restore from a backup on deploy.

Group 18 — Custom SQL Scripts

enable_custom_sql_scripts, custom_sql_scripts_bucket, custom_sql_scripts_path, custom_sql_scripts_use_root — run SQL from a GCS bucket after provisioning. See App_GKE.

Group 19 — Custom Domain, Static IP & Networking

Variable	Default	Description
enable_custom_domain	true	Provision Kubernetes Gateway API for custom hostnames + managed certificate.
application_domains	[]	Hostnames to serve.
reserve_static_ip	true	Stable external IP across redeploys.

Group 20 — Identity-Aware Proxy (IAP)

Variable	Default	Description
enable_iap	false	Require Google sign-in in front of n8n. Note: enabling IAP blocks public webhook endpoints.
iap_authorized_users / iap_authorized_groups	[]	Who may access.
iap_oauth_client_id / iap_oauth_client_secret	""	Required when IAP is enabled (sensitive).

Group 21 — Cloud Armor & CDN

Variable	Default	Description
enable_cloud_armor	false	Attach a Cloud Armor (WAF) policy to the Ingress backend.
admin_ip_ranges	[]	CIDRs allowed privileged access.
cloud_armor_policy_name	(set)	Policy name.
enable_cdn	false	Enable Cloud CDN on the backend.

Group 22 — VPC Service Controls & Audit Logging

Variable	Default	Description
enable_vpc_sc	false	Enforce a VPC-SC perimeter (requires organization_id).
vpc_cidr_ranges / vpc_sc_dry_run	(set)	Access level CIDRs / dry-run mode.
enable_audit_logging	false	Detailed Cloud Audit Logs.

Group — AI Components

Variable	Default	Description
enable_ai_components	true	Master toggle. Set false to deploy n8n without Qdrant or Ollama.
enable_qdrant	true	Deploy Qdrant as an internal-only Kubernetes Deployment.
qdrant_version	latest	Qdrant Docker image tag. Pin to a specific version for production stability.
enable_ollama	true	Deploy Ollama as an internal-only Kubernetes Deployment.
ollama_version	latest	Ollama Docker image tag. Pin to a specific version for production stability.
ollama_model	llama3.2	Default model declaration. Note: this variable is not currently forwarded to the Ollama service — models must be pulled separately at the Ollama API level.

---

5. Outputs

These values are returned on a successful deployment and are the quickest way to locate and explore the running resources.

Output	Description
service_name	Kubernetes Service name for the n8n workload.
namespace	Namespace all workloads (n8n, Qdrant, Ollama) run in.
service_cluster_ip	In-cluster ClusterIP for the n8n Service.
stage_service_cluster_ips	Map of ClusterIPs for stage-specific services (Cloud Deploy).
service_external_ip	External LoadBalancer IP (when a static IP is reserved).
service_url	URL to reach the n8n UI.
database_instance_name	Cloud SQL instance name.
database_name	Application database name (n8n_db).
database_user	Application database user (n8n_user).
database_password_secret	Secret Manager secret holding the DB password.
database_host / database_port	DB endpoint (127.0.0.1 via Auth Proxy) / port.
storage_buckets	Created Cloud Storage buckets (including the AI data bucket).
network_name / network_exists / regions	VPC network, presence, available regions.
container_image / container_registry	Deployed image and Artifact Registry repo.
monitoring_enabled / monitoring_notification_channels	Monitoring status and channels.
initialization_jobs / db_import_job	Names of the setup and (optional) import jobs.
deployment_id / tenant_id / resource_prefix	Naming identifiers.
project_id / project_number	Project identifiers.
cicd_enabled / cicd_configuration	CI/CD status and details (repo, trigger, registry).
github_repository_url / github_repository_owner / github_repository_name	Connected repo details.
artifact_registry_repository / cloudbuild_trigger_name / cloudbuild_trigger_id	Registry and build trigger.
kubernetes_ready	Whether the cluster/workload is ready.
vpc_sc_enabled / vpc_sc_perimeter_name / vpc_sc_dry_run_mode	VPC-SC status.
audit_logging_enabled / artifact_registry_cmek_enabled	Audit logging and CMEK status.

---

6. Configuration Pitfalls & Sensible Defaults

Risk: Critical (data loss / outage / security) — High (service degraded) — Medium (cost or partial degradation) — Low (minor).

Setting	Sensible value	Risk	Consequence if wrong
N8N_ENCRYPTION_KEY (auto-generated)	Back up immediately	Critical	Changing after first run permanently destroys all saved n8n credentials.
application_name	n8nai — set once	Critical	Immutable after first deploy; renaming recreates all GCP and Kubernetes resources with data loss.
db_name / db_user	set once	Critical	Immutable after first deploy; renaming points n8n at a new empty database, losing all workflows.
enable_backup_import	false unless restoring	Critical	Enabling without a valid backup_uri fails the import job.
quota_memory_requests / _limits	binary units	Critical	Bare integers are bytes and block all pod scheduling.
enable_qdrant	true	High	Active RAG workflows fail at runtime with connection errors if Qdrant is removed.
enable_ollama	true	High	Workflows using the local LLM node fail; only disable when using external AI providers exclusively.
enable_redis	true	High	Without Redis, multiple replicas conflict on workflow state; split-brain execution corrupts runs.
redis_host	"" (NFS) or explicit	High	When Redis is on but both redis_host and NFS are unset, n8n fails to start.
memory_limit	4Gi	High	AI workflows (embedding, vector search, LLM chaining) cause OOM kills below 4 GiB.
max_instance_count	1 unless Redis configured	High	Scaling above 1 without Redis causes split-brain; increasing with Redis is safe.
min_instance_count	1 for webhooks	Medium	0 may leave webhooks without a target pod; HPA state can be inconsistent on GKE.
enable_nfs	true	High	Qdrant and Ollama use GCS Fuse on the AI data bucket; without it model files and vector indexes are lost on pod restart.
enable_iap	only with valid OAuth creds	High	Enabling without iap_oauth_client_id / iap_oauth_client_secret blocks all access. IAP also blocks public webhooks.
backup_retention_days	7 (raise for prod)	Medium	Too short for compliance retention.
pdb_min_available vs min_instance_count	leave headroom	Medium	1/1 can stall node upgrades (single pod cannot be evicted).

---

For the foundation behaviour referenced throughout — IAM and Workload Identity, autoscaling, ingress and certificates, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC-SC, backups, and image mirroring — see App_GKE. n8n AI-specific application configuration shared with the Cloud Run variant is described in N8N_AI_Common.