// product · available now
n8n Production Self-Hosting Kit
hardened Compose, queue mode, backups & Helm
Self-host n8n properly: hardened Docker Compose stacks, queue mode with task runners, backups that are proven to restore, Prometheus alerts and a Helm chart.
● instant download · secure checkout by Gumroad
// tested
Tested, not just written
Every release runs its full test suite against real tooling before it ships.
A webhook through Caddy over HTTPS runs end to end: Code node in the external runner, credential decrypted, expected JSON back
Stopping the runner breaks Code nodes and starting it fixes them, so the runner really is external
backup → destroy the volumes → restore: the workflow still exists, is published, executes, and its credential decrypts
A backup restored with the wrong encryption key is refused before anything is touched (exit code 3)
Queue mode: webhook processors serve, workers execute, and a killed worker leaves its jobs to the survivor
32 dashboard panel queries return data, 975 scraped metric names still exist, and stopping the workers fires the alert
Studio: the chart installs on a kind cluster under Pod Security Admission restricted, with 13 assertions and helm test
// pricing
One-time purchase, three tiers
Pay once, keep the files. Each tier includes everything in the tier before it.
Starter
Teams running n8n on one server
$19USD
- Hardened Compose stack: n8n, external task runner, PostgreSQL 17, Caddy with automatic HTTPS
- Non-root, read-only filesystems, internal network, resource limits, healthchecks
- init-secrets.sh, backup.sh and restore.sh with key fingerprinting and checksums
- Install guide, upgrade runbook, security checklist, environment reference
- 30 days of updates
Pro
Teams scaling n8n past one instance
$49USD
- Queue mode: main, webhook processors, worker + task runner pairs, Valkey
- Caddy routes /webhook/* to the processors and re-resolves DNS, so scaling needs no reload
- Prometheus scrape config, 16 alert rules with promtool unit tests, 20-panel Grafana dashboard
- Scheduled backups with verify-backup.sh restoring into throwaway containers
- Scaling, logging and backup guides, 1 year of updates
Studio
Kubernetes platform teams, consultants, agencies
$99USD
- ftw-n8n Helm chart: main, webhook and worker Deployments, runner sidecars, HPA, PDBs, NetworkPolicies
- values.schema.json, helm test hook, dashboard ConfigMap, CRD-gated ServiceMonitor and PrometheusRule
- Workflow promotion with Git: export/import scripts and a GitHub Actions example
- Kubernetes sizing guide, client delivery checklist, client-use license
- 1 year of updates
// contents
What's inside
- Hardened single-instance Compose stackStarter+
- Caddy with automatic HTTPSStarter+
- External task runnerStarter+
- Backup, restore & secret generationStarter+
- Queue mode: main, webhooks, workersPro+
- 16 alert rules + promtool testsPro+
- 20-panel Grafana dashboardPro+
- Scheduled + verified backupsPro+
- Scaling, logging & backup guidesPro+
- ftw-n8n Helm chartStudio+
- Git workflow promotion + CI exampleStudio+
- Kubernetes sizing & delivery checklistStudio+
// compatibility
Works with your stack
- n8n 2.39.7 + n8n task runners 2.39.7 (pinned everywhere)
- PostgreSQL 17.11 · Valkey 8.1.10 · Caddy 2.11.4
- Docker Compose v2 on any Linux host · amd64 and arm64
- Prometheus 3.14 · Grafana 13.2
- Kubernetes 1.32+ with Helm 3.14+ / 4 (Studio)
// faq
Questions
+Why not just run the docker run command from the docs?
It works until it doesn't: SQLite locking, no backups, executions lost on restart, no HTTPS, no metrics. This kit is the version you can still operate a year in, with a restore path that is tested rather than assumed.
+Do I need queue mode?
Not at first. Starter is a single hardened instance. Move to Pro's queue mode when executions queue behind each other, webhooks need to stay responsive, or you want to scale workers independently.
+Does this include n8n itself?
No. It is deployment configuration, scripts, monitoring and documentation for running the official n8n images, which stay under n8n's own Sustainable Use License. The Studio client-use license covers our files only.
+Can I use my existing Postgres and Redis?
Yes, and that is the production default in the Helm chart. The in-chart datastores are marked evaluation only.
+Can I use it for client work?
Starter and Pro are licensed for your own organization. Client and agency use requires Studio.
+Refunds?
If the kit doesn't work with the pinned n8n version and the troubleshooting guide doesn't fix it, contact us within 14 days for a refund.
Run n8n like you mean it.
Questions first? Email hello@fractaltechware.com