DEV Community: Vikas Singhal

How to Monitor Proxmox with Beszel in 5 Minutes (2026)

Vikas Singhal — Sat, 23 May 2026 10:27:06 +0000

To monitor a Proxmox VE host with Beszel: install the Beszel Hub on any internet-reachable server, then run the Beszel agent on the Proxmox host with a one-liner installer (curl -sL https://get.beszel.dev | bash). The agent uses about 10MB of RAM, exposes port 45876, and reports CPU, memory, disk, network, temperature, and Docker container metrics to the Hub. No license. No telemetry. No per-host pricing.

I run Beszel across two Proxmox hosts and a handful of LXC containers in my homelab. Total monthly cost: $0 (Hub on a free-tier pod) or about $3/mo if you want a managed Hub. Total RAM footprint per machine: 10MB. Total time to set up the first host: under 5 minutes once the Hub is running.

This guide walks through the full setup, including the LXC-vs-host decision, the systemd service config that survives Proxmox reboots, and the gotchas that trip people up.

What is Beszel?

Beszel is an open-source server monitoring tool built by Henry Doss. It has two parts:

Beszel Hub: a web dashboard you self-host once. It receives metrics from every agent and renders the charts, alerts, and history.
Beszel Agent: a tiny binary you install on each machine you want to monitor. It reports CPU, RAM, disk usage, network throughput, swap, temperature, and Docker container stats.

The agent uses about 10MB of RAM. For comparison: Netdata's agent uses 200-500MB. Datadog's agent uses 500MB+ and starts at $15/host/month. Beszel is free and self-hosted.

For Proxmox VE specifically, Beszel gives you per-host load, memory pressure, ZFS pool usage, and a per-container view if you run Docker on the host. It does not replace the Proxmox cluster manager (you still want that for backups and migrations), but it gives you the historical graphs and alert hooks Proxmox itself does not.

Step-by-Step: Install Beszel on Proxmox

The setup has two phases. Phase 1 is one-time. Phase 2 you repeat for every Proxmox host or LXC container you want to monitor.

Phase 1: Set up the Beszel Hub

The Hub needs to live somewhere reachable from your Proxmox hosts. Options:

A cheap VPS (Hetzner CX11 at $4/mo, InstaPods Launch plan at $3/mo, or any small Linux box).
An LXC container on your Proxmox host itself (free, but you lose monitoring if the host goes down).
A Raspberry Pi on the same network (zero cloud cost, but no remote alerts).

For most homelabs I recommend option 1 or 3. If you only have one Proxmox host, option 1 wins because you can still see metrics when the host reboots.

For a one-click Hub on a $3 pod with HTTPS handled: deploy the Beszel app on InstaPods. Otherwise, pull the official Docker image:

docker run -d \
  --name beszel \
  --restart unless-stopped \
  -p 8090:8090 \
  -v ./beszel_data:/beszel_data \
  henrygd/beszel

Open http://<hub-host>:8090/ in a browser. The first request prompts you to create an admin account. Save the URL - the agents need it.

Phase 2: Install the Beszel Agent on the Proxmox host

SSH into the Proxmox host as root:

ssh root@your-proxmox-host

In the Beszel Hub web UI, click Add System. Copy the public key it generates - the agent needs this to authenticate.

Run the installer one-liner:

curl -sL https://get.beszel.dev -o /tmp/install-agent.sh && \
  chmod +x /tmp/install-agent.sh && \
  /tmp/install-agent.sh

The installer prompts for:

Public key: paste from the Hub
Port: leave default (45876)
Hub URL (optional): if your Hub is on a different network, paste the full URL
Token (optional): only needed for advanced multi-tenant setups

The installer creates a systemd service at /etc/systemd/system/beszel-agent.service and starts it. Verify:

systemctl status beszel-agent

Output should show active (running). Within 30 seconds, the host appears in the Beszel Hub dashboard with live charts.

Phase 3 (Optional): Monitor LXC Containers Individually

Proxmox LXC containers do not appear automatically as separate systems in Beszel - they share the host's namespace. If you want each LXC monitored independently, install the agent inside the container:

pct enter <container-id>
curl -sL https://get.beszel.dev | bash
exit

For privileged LXC containers this works directly. For unprivileged containers you may need to enable the mount=cgroup feature in the container's .conf file at /etc/pve/lxc/<id>.conf. Most monitoring metrics (CPU, RAM, network) work without this; full Docker stats inside the LXC need it.

Phase 4: Make it Reboot-Proof

The installer already enables the service via systemctl enable beszel-agent.service. To verify after a reboot:

sudo reboot
# wait, SSH back in
systemctl is-enabled beszel-agent
# expected: enabled
systemctl is-active beszel-agent
# expected: active

If the agent does not come up, the most common cause is firewall. Check that port 45876 is open between the host and the Hub:

iptables -L | grep 45876
# or for nftables
nft list ruleset | grep 45876

Beszel vs Other Proxmox Monitoring Options

Tool	RAM per agent	Setup time	Cost	Best for
Beszel	10 MB	5 min	$0 (self-hosted)	Homelabs, small fleets, Proxmox + Docker
Netdata	200-500 MB	10 min	$0 (free tier) or $69/mo (cloud)	Granular per-second metrics, ML anomaly detection
Datadog Agent	500+ MB	30 min	$15+/host/month	Large enterprise fleets, compliance
Grafana + Prometheus + Node Exporter	200+ MB stack	60+ min	$0 (self-hosted), complex	Customizable dashboards, alerting flexibility
Proxmox built-in graphs	0 (already there)	0 min	$0	Single-host CPU/RAM at-a-glance only
Zabbix	50-200 MB	60+ min	$0 (self-hosted), enterprise pricing	Network device monitoring, large fleets

For most Proxmox homelabs running 1-10 hosts, Beszel hits the sweet spot. It is more lightweight than Netdata, simpler than Grafana + Prometheus, and free vs Datadog. The trade-off is fewer metrics types - if you need per-second granularity or ML anomaly detection, Netdata is the better pick.

Frequently Asked Questions

Does Beszel work on Proxmox VE 8?

Yes. The agent is a standalone Linux binary that runs on any modern systemd-based Linux distribution. Proxmox VE 7 and 8 (both Debian-based) work without modification.

Will the Beszel agent interfere with Proxmox's own monitoring?

No. Beszel reads from /proc and /sys, the same places Proxmox's built-in graphs read from. There is no conflict and no measurable performance overhead.

Can I monitor ZFS pools with Beszel?

Beszel reports disk usage for mounted filesystems including ZFS. It does not report ZFS-specific metrics like ARC hit ratio or scrub status. For those, pair Beszel with zpool status cron alerts or use zfs-mon.

How do I add alerts when a Proxmox host goes down?

In the Beszel Hub, open any system and click the bell icon next to the metric you want to alert on. Beszel supports email, Discord, Slack, ntfy, Pushover, Gotify, and webhook notifications. The Hub watches for missing agent heartbeats and fires a "system down" alert automatically.

What if I run Proxmox in a cluster?

Install the agent on each cluster node. Each node reports independently to the Hub. The Hub does not currently aggregate cluster-level views (per-cluster RAM, per-cluster VM count), so you get a per-node dashboard rather than a cluster overview.

Is there a Beszel agent for Windows Server?

Yes, as of Beszel 0.8+. Same install pattern with a PowerShell installer instead of bash. Most Proxmox users do not need this, but it is there if you have a mixed-OS homelab.

Cost Comparison: One Year of Proxmox Monitoring

For a 3-host Proxmox cluster monitored over 12 months:

Setup	Year 1 cost
Beszel + Hub on InstaPods $3/mo Launch plan	$36
Beszel + Hub on Hetzner CX11 €4/mo	~$56
Beszel + Hub self-hosted on existing Pi/VPS	$0
Netdata Cloud Pro (3 hosts)	$99
Datadog Pro (3 hosts)	$540

Beszel pays for itself the first month if you were paying for Netdata or Datadog. The license is open source. There is no per-host or per-metric pricing.

Wrapping Up

Beszel on Proxmox is the lightest-weight serious monitoring you can get. 10MB of RAM per host, a real web UI with charts and alerts, and Docker container visibility for free. The setup is two phases - Hub once, agent per host - and the agent ships its own systemd service so it survives reboots.

If you want zero-config Hub hosting with HTTPS and a real backup story, Beszel runs as a 1-click app on InstaPods for $3/mo. You can also see how it stacks against Netdata, Datadog, and Prometheus in our comparison cluster.

What does your Proxmox monitoring stack look like? Drop a comment - I'm especially curious about anyone running Beszel alongside Netdata or Grafana to see if they're worth the combined overhead.

I Tested 7 Self-Hosted Monitoring Tools on a $3 VPS in 2026 (Here's the One I Kept)

Vikas Singhal — Sat, 16 May 2026 03:38:40 +0000

I was paying around $84/year for UptimeRobot's Pro plan plus another $20/month for a Datadog free-tier-with-overage thing on a side project. None of it made sense. The side project does maybe 200 requests an hour. I'm not running Netflix.

So over a long weekend in 2026 I spun up a $3/mo VPS, installed 7 self-hosted monitoring tools one after the other on the same box, and kept whichever one didn't make me reach for rm -rf at the end.

Same hardware for all of them: 1 vCPU, 2 GB RAM, 25 GB disk, Ubuntu 24.04. Same workload: 4 production websites pinged from outside + the host itself.

The 7 tools (and how I ranked them)

Tool	Memory at idle	Setup time	First-time UX	Notes
Beszel	12 MB (hub) + 8 MB (agent)	4 min	"Wait, this is it?"	Single binary, agent + hub. Zero config.
Netdata	320 MB	6 min	Overwhelming	Auto-detects everything. So many charts.
Grafana + Prometheus	410 MB combined	35 min	Painful	You install one to be useful, the other to make the first one useful.
Pulse (Proxmox VE)	95 MB	8 min	Niche	Lovely if you run Proxmox. Empty otherwise.
Glances + InfluxDB	240 MB	20 min	Average	Glances is fine. InfluxDB is the tax.
Prometheus only	180 MB	15 min	Terminal-grade	You're querying PromQL or you're nothing.
Dozzle	18 MB	3 min	Logs only	Brilliant at one thing. Not a monitoring tool.

The memory column matters more than it looks. On a 2 GB VPS, Netdata's 320 MB is 16% of your RAM gone before you've shipped any user code. Beszel's 12 MB is a rounding error.

Why I uninstalled the obvious choices first

Netdata. Auto-detects PostgreSQL, nginx, Docker, journalctl, every interface, every cgroup. Great. Also: 320 MB at idle and a dashboard so dense I scrolled for 90 seconds and gave up. It's the Salesforce of self-hosted monitoring - extremely capable, exhausting to use casually.

Grafana + Prometheus. I love Grafana. I do not love that the answer to "how is my server doing" is "let me first configure scrape targets in a YAML file, then write a PromQL query, then build a dashboard, then realize the dashboard library on GrafanaLabs has 600 versions of CPU usage." 35 minutes to first chart. On a side project.

Pulse. I don't run Proxmox. It told me so within 90 seconds. Fair.

Glances. Glances itself is great as a CLI. The moment you want web UI + history + alerts, you need InfluxDB or some other time-series store, and now you've got a moving-parts problem on a 2 GB box.

Dozzle. Spectacular for tailing Docker logs. Doesn't pretend to be monitoring.

Prometheus alone. PromQL is wonderful if you live in it. I don't. I want to glance at my phone in the queue at the coffee shop and know whether prod is up.

The one I kept: Beszel

Beszel is two binaries:

The hub runs on one server. Web UI, history, alerts, push notifications.
The agent runs on every server you want to watch. Reports back to the hub over an encrypted channel.

Setup, end to end:

# On the hub server
mkdir beszel && cd beszel
curl -L https://github.com/henrygd/beszel/releases/latest/download/beszel_Linux_x86_64.tar.gz \
  | tar -xz
./beszel serve

# On every agent server
curl -L https://github.com/henrygd/beszel/releases/latest/download/beszel-agent_Linux_x86_64.tar.gz \
  | tar -xz
KEY="$(cat ~/.ssh/id_ed25519.pub)" ./beszel-agent

# Add the agent in the hub web UI, paste the public key, done

That's 4 minutes if you've made a single typo, 3 minutes if you haven't.

The web UI shows CPU, memory, disk, network, temperature, Docker container stats, and a clean event log. Alerts go to email, Discord, ntfy, Pushover, Telegram - the usual stack. History as far back as you tell it to keep.

What sealed it for me: the same setup works for 1 server or 50. I started with the one VPS, then added the agent to my home homelab box (an old laptop running Debian under the TV), then to a Raspberry Pi running Home Assistant. Same dashboard. Same alerting. No reconfiguration.

What I'd actually pay attention to

Three things matter on a small server: memory headroom, disk filling up overnight, and "did the process I care about die." Beszel covers all three out of the box. Netdata covers them too, but you'll need to learn the dashboard. Grafana covers them once you've spent a weekend wiring it together.

If you run multiple boxes - even just "VPS + home lab + Pi" - the single-hub model is the unlock.

Costs (real numbers)

Setup	Monthly cost	What you give up
UptimeRobot Pro	$7	External-only checks, no resource metrics
Datadog Pro	$15+ per host	Per-metric overage. The trap.
Grafana Cloud Free	$0	10K metrics ceiling, then it scales fast
Beszel on a $3 VPS	$3	Hosting it yourself, owning your data, no per-host fee
Beszel on a managed pod	$3	Setting it up yourself

The last row is shameless - I work on InstaPods and we ship a pre-baked Beszel image for people who don't want to run the install commands above. If you'd rather not touch a VPS, that path exists. If you'd rather own the box, the install commands above are the whole guide.

What I'd skip

Don't self-host monitoring on the same box you're monitoring if it's a one-server side project. If that box goes down, your monitor goes down with it. Beszel's hub is small enough to run on a different cheap VPS, a Raspberry Pi at home, or a free-tier Oracle box.
Don't pay for SaaS monitoring on a side project. The pricing model assumes you're running 10 production hosts and have a budget. You're running one production host and a budget of "ramen tonight or not." The math never works.
Don't pick the tool with the prettiest dashboard until you've tried the tool with the smallest memory footprint. The prettier dashboard you can grow into. The 320 MB you can't grow into on a 2 GB box.

Try-this list (15 minutes total)

Spin up the cheapest VPS you can find. Beszel's hub needs maybe 50 MB and a few MB of disk.
Install Beszel hub. Open the web UI.
Install the agent on whatever server you care about. Paste the SSH public key into the hub. Done.
Set one alert: "CPU > 90% for 5 minutes." That's the single alert that catches the most "something is wrong" cases.

The whole thing costs less than one month of UptimeRobot Pro and tells you more.

What's your monitoring stack right now? And how many alerts do you actually act on vs. silently ignore? Curious how this lands with people running multi-box homelabs - the agent/hub model seems made for it but I've only run it on three boxes so far.

Why I Stopped Paying for n8n Cloud (And Switched to a $4 VPS in 2026)

Vikas Singhal — Thu, 14 May 2026 02:31:10 +0000

I was paying n8n Cloud $20 a month for the Pro tier. Eight workflows, mostly daily syncs. Nothing crazy.

Then I looked at the execution counter. I was at 14,200 executions for the month. Pro caps at 10,000. Next tier up: $50 a month. For workflows that mostly just call APIs and write to a database.

So I spent a Sunday afternoon migrating to self-hosted. Here's what I learned.

n8n Cloud pricing in 2026 (the math nobody shows you)

n8n Cloud's pricing is execution-based, not user-based. Real numbers from their site:

Tier	Price	Executions	Active workflows
Starter	$20/mo	2,500	5
Pro	$50/mo	10,000	15
Enterprise	Custom	Custom	Unlimited

The catch is that "executions" counts every workflow run. A daily sync that runs 30 times a month uses 30 executions. A webhook-triggered workflow that fires 50 times a day uses 1,500. Eight workflows running a few times a day each adds up faster than you'd think.

If you have AI-heavy workflows (LLM calls, embeddings, vector search loops), executions stack even faster. Each LangChain agent step counts.

I crunched my own numbers: at Pro tier ($50/mo, 10K executions), I'd burn through the quota in 22 days. The next overage tier was disproportionate. Self-hosted made sense.

The self-host options in 2026

Three real paths:

1. Raw VPS (Hetzner CX22 at $4.51/mo)

You rent a 2 vCPU / 2GB RAM box from Hetzner, install Docker, and run n8n with Docker Compose:

services:
  n8n:
    image: n8nio/n8n
    ports:
      - "5678:5678"
    environment:
      - N8N_HOST=n8n.yourdomain.com
      - WEBHOOK_URL=https://n8n.yourdomain.com
    volumes:
      - ~/.n8n:/home/node/.n8n

Then nginx in front for SSL, certbot for a Let's Encrypt cert, fail2ban for SSH, ufw for firewall, automated backups for the data volume. Maybe 90 minutes of setup if you've done it before. Multiple hours if it's your first time.

Total monthly cost: $4.51 for the VPS. You handle everything else.

2. Managed PaaS (Coolify, Dokploy, InstaPods, Sevalla)

These sit between raw VPS and n8n Cloud. You get a control panel that handles SSL, backups, and updates, but the n8n instance is yours. Pricing is flat, not execution-based.

I tried Coolify (self-installed on the same Hetzner VPS) and InstaPods (which has a pre-baked 1-click n8n image). The InstaPods approach was faster because the image is pre-configured: deploy n8n in under a minute, SSL auto-provisioned, terminal access in the browser. $3/mo on their Launch tier.

The Coolify approach takes longer to set up but you own the entire stack. Both are valid; the choice depends on whether you want to manage the platform itself.

3. Docker on a Raspberry Pi or home server

If you already have a Pi 4 or a homelab box, you can run n8n on hardware you own. Zero monthly cost beyond electricity. The trade-off is exposing it: you need Cloudflare Tunnel, Tailscale, or a public IP plus dynamic DNS.

I considered this but my Pi was already running Plex and Home Assistant. Adding n8n was one container too many for the SD card.

What I actually do now

I'm on a $3/mo InstaPods pod with n8n pre-installed. Total monthly bill: $3. That's a savings of $17/month versus n8n Cloud Starter, or $47/month versus Pro.

The workflows that mattered:

Stripe webhook -> Postgres -> Slack (30-50 executions/day)
RSS feed monitor -> Discord (4 daily)
GitHub issue triage -> Linear (5-20 per day depending on day)
OpenAI summarization of long emails -> Notion (10-30 daily)

None of them needed n8n Cloud's SLA. The whole point of self-hosting was: I control the database, I control the credentials, I can SSH in and debug, and the bill is fixed.

Migration took 90 minutes total. Export workflows as JSON from n8n Cloud, import into self-hosted, re-add credentials (n8n encrypts them, so they don't export). Done.

When NOT to self-host

To be fair, self-hosting isn't always the right call:

If your workflows handle critical production traffic with strict uptime requirements, n8n Cloud's SLA is worth the premium.
If your team isn't comfortable with Linux/Docker, the setup time eats the cost savings.
If you're below 2,500 executions/month forever, the $20 Starter tier might genuinely be cheaper than your time.

Self-hosting wins clearly above 5,000 executions/month or when you want to own the data.

Discussion

Are you self-hosting n8n right now? What was the breaking point that made you migrate (or kept you on Cloud)? Curious how others crunch the math.

I Replaced Datadog With a 10MB Monitoring Tool (Here's What Happened)

Vikas Singhal — Fri, 08 May 2026 06:43:35 +0000

I have 4 servers. Nothing enterprise-grade - a couple of app servers, a database box, and a staging machine. Datadog was costing me $60/month to show me CPU graphs and disk usage.

That felt wrong.

So I tested alternatives. Grafana + Prometheus was the obvious choice, but the stack itself uses 500MB+ of RAM. On a $4/mo VPS with 2GB total, that's 25% of my resources just for monitoring.

Then I found Beszel. Its agent uses less than 10MB of RAM. The entire monitoring stack - hub and agents across 4 servers - runs on a $3/mo server and barely touches 50MB.

Here's what happened when I switched.

What Beszel Actually Monitors

Beszel tracks the basics well:

CPU usage and load average
Memory and swap usage
Disk usage and I/O
Network throughput
Docker container metrics (CPU/memory per container)

It does NOT do:

Application Performance Monitoring (APM)
Distributed tracing
Log aggregation
Custom metrics/dashboards beyond server basics
Alerting with complex conditions (basic alerts only)

If you need APM and tracing, Beszel isn't for you. If you need "is my server healthy?" at a glance, it's all you need.

The Setup (5 Minutes)

Hub (runs on one server - I used a $3/mo InstaPods server):

The hub is the web dashboard that collects and displays metrics. One-click deploy gives you a running instance with HTTPS in about 60 seconds.

Agent (runs on each monitored server):

curl -sL https://raw.githubusercontent.com/henrygd/beszel/main/supplemental/scripts/install-agent.sh -o install-agent.sh
chmod +x install-agent.sh
./install-agent.sh

Enter your hub URL when prompted. The agent starts reporting metrics immediately.

Total setup time across 4 servers: about 5 minutes. Compare that to Prometheus + Grafana (30-60 minutes for the stack, then configuring dashboards, then setting up node_exporter on each server).

Resource Usage Comparison

I measured actual resource usage across monitoring tools on identical VPS instances (2 vCPU, 4GB RAM):

Tool	Agent RAM	Hub/Server RAM	Total (4 agents + hub)
Beszel	<10MB	~50MB	~90MB
Netdata	200-500MB	N/A (each instance is standalone)	800-2000MB
Prometheus + node_exporter	50-100MB (exporter)	300-500MB (Prometheus)	500-900MB
Grafana (if you add it)	N/A	200-400MB	adds 200-400MB on top
Datadog agent	200-500MB	N/A (cloud)	800-2000MB

Beszel's total footprint across my entire infrastructure (hub + 4 agents) is less than what a single Netdata or Datadog agent uses on one server.

What I Lost

Switching from Datadog means losing:

APM traces - I can't see request latency broken down by endpoint. For debugging slow API calls, I now SSH in and check logs manually.
Log search - Datadog's log search was convenient. I now use grep and journalctl on the servers directly.
Custom dashboards - Datadog lets you build anything. Beszel's dashboard is fixed (server metrics only).
Integrations - Datadog connects to everything (AWS, Kubernetes, databases). Beszel monitors the server, not specific services.
Alerts - Datadog's alerting is sophisticated (anomaly detection, composite conditions). Beszel has basic threshold alerts.

For my 4-server setup, none of these losses mattered. I wasn't using APM. My logs are small enough to grep. My dashboard needs are basic.

What I Gained

$57/month savings - $60/month (Datadog) vs $3/month (Beszel hub hosting). That's $684/year.
No data leaving my infrastructure - monitoring data stays on my servers. No third-party data processor.
Lower server overhead - freeing up 200-500MB per server means my apps have more headroom.
Simpler mental model - one dashboard, server metrics, done. No 47-tab Datadog interface.
No per-host pricing anxiety - adding a 5th server to Beszel costs $0 extra. Adding it to Datadog costs $15/month.

When Beszel Is Wrong

Don't switch to Beszel if:

You're running 50+ servers (you need the scalability of Prometheus/Grafana or a cloud solution)
You need APM with distributed tracing (use Datadog, New Relic, or self-hosted Jaeger)
You need log aggregation (use Loki, Elasticsearch, or a cloud logging service)
Your compliance requires specific monitoring certifications
You have a team of SREs who need advanced dashboards and runbooks

When Beszel Is Right

Switch to Beszel if:

You're monitoring 1-20 servers
You need CPU, memory, disk, network metrics
You want Docker container monitoring
Budget matters (your credit card, not the company's)
You value simplicity over features
You want to self-host your monitoring data

My Setup Now

Beszel hub: $3/mo managed server (InstaPods, one-click deploy)
4 agents: installed in ~1 minute each, <10MB RAM each
Total cost: $3/month
What I check daily: CPU load, memory %, disk space, network throughput
Time saved vs Datadog: I spend less time in the monitoring UI because there's less to get distracted by

The honest answer: Beszel gives me 80% of what I used Datadog for at 5% of the cost. The other 20% (APM, logs, custom dashboards) I handle with SSH and grep when I need them, which is rarely.

What monitoring tool are you running? Curious if anyone else has downsized from Datadog/New Relic to something lighter - and whether you regret it.

The Cheapest Way to Self-Host Beszel in 2026

Vikas Singhal — Fri, 24 Apr 2026 06:23:42 +0000

Datadog charges $15/host/month. You have 5 servers. That's $75/month to see CPU graphs.

Beszel does the same job with an agent that uses less than 10MB of RAM. The entire monitoring stack - hub + agent - runs comfortably on the cheapest server you can buy.

I've been running Beszel across 4 servers for the past few months. Here's every way to host it in 2026, with what each option actually costs.

What Is Beszel?

Beszel is a lightweight, self-hosted server monitoring tool. It has two components:

Hub - the web dashboard that collects and displays metrics (runs on one server)
Agent - a tiny binary that runs on each monitored server and reports metrics back

The agent uses <10MB of RAM. Compare that to:

Netdata agent: 200-500MB
Datadog agent: 200-500MB
Prometheus + node_exporter: 100-300MB
Full Grafana stack: 500MB+

For small to medium infrastructure (1-20 servers), Beszel gives you CPU, memory, disk, network, and Docker container metrics with a clean dashboard. No YAML configs, no query languages, no data retention costs.

Method 1: One-Click Managed Hosting ($3/mo)

Setup time: 60 seconds
Monthly cost: $3
Best for: Getting Beszel running instantly without managing a server

Managed hosting platforms deploy Beszel pre-configured. You get the hub running at a live URL with HTTPS in about a minute. Then you install the agent on each server you want to monitor.

The hub is lightweight enough to run on a $3/mo plan (0.5 vCPU, 512MB RAM). I've been running it with 4 agents reporting every 10 seconds and it barely touches 50MB of RAM.

On InstaPods, the deploy takes one click. You get SSH access if you need to customize the config, automatic HTTPS, and no server management.

After the hub is running, install the agent on each server you want to monitor:

curl -sL https://raw.githubusercontent.com/henrygd/beszel/main/supplemental/scripts/install-agent.sh -o install-agent.sh && chmod +x install-agent.sh && ./install-agent.sh

The agent connects to your hub URL and starts reporting metrics immediately.

Pros: Fastest setup, zero maintenance, cheapest option
Cons: Hub runs on managed infra (you SSH in but don't control the OS)

Method 2: Docker on a VPS ($4-5/mo)

Setup time: 15-30 minutes
Monthly cost: $4-5
Best for: Developers who want full control

Rent a VPS from Hetzner ($4.51/mo for CX22) or DigitalOcean ($4/mo), install Docker, and run Beszel with Docker Compose:

services:
  beszel:
    image: henrygd/beszel
    container_name: beszel
    restart: unless-stopped
    ports:
      - 8090:8090
    volumes:
      - ./beszel_data:/beszel_data

Then add nginx as a reverse proxy and set up Let's Encrypt for HTTPS.

Total setup: about 15 minutes if you've done Docker deploys before. 30+ minutes if it's your first time.

Pros: Full control, can run other services on the same VPS
Cons: You manage everything - updates, SSL renewal, backups, nginx config

Method 3: Run It on an Existing Server ($0)

Setup time: 5 minutes
Monthly cost: $0 (uses existing resources)
Best for: If you already have a VPS running other services

Beszel is light enough to add to any existing server without impacting performance. The hub uses ~50MB RAM under normal load. Install it directly:

curl -sL https://raw.githubusercontent.com/henrygd/beszel/main/supplemental/scripts/install-hub.sh | bash

This installs the hub binary and sets it up as a systemd service. No Docker required. It runs on port 8090 by default.

If you already have a reverse proxy (nginx/caddy) running on that server, add a server block to expose Beszel on a subdomain.

Pros: Free, minimal resource usage, no Docker needed
Cons: Shares resources with other services, monitoring your monitor (if the server goes down, you lose visibility)

Method 4: Elestio (~$17/mo)

Setup time: 5 minutes
Monthly cost: ~$17
Best for: Teams that want fully managed hosting with support

Elestio handles deployment, updates, backups, and monitoring of your Beszel instance. You get enterprise-grade hosting without touching a terminal.

At $17/mo for a monitoring tool, the cost only makes sense for teams managing production infrastructure where uptime matters enough to justify the premium.

Pros: Fully managed, includes backups and monitoring
Cons: Most expensive option by far for a monitoring tool

Cost Comparison

Method	Monthly Cost	Setup Time	You Manage	Docker Required
Managed hosting	$3	60 sec	Nothing	No
Docker on VPS	$4-5	15-30 min	Everything	Yes
Existing server	$0	5 min	Beszel updates	No
Elestio	$17	5 min	Nothing	No

Beszel vs Datadog/New Relic: When to Use What

Beszel replaces Datadog and New Relic for a specific use case: server-level monitoring for small to medium infrastructure.

Use Beszel when:

You're monitoring 1-20 servers
You need CPU, memory, disk, network metrics
You want Docker container monitoring
You don't need APM (application performance monitoring)
Budget matters more than features

Use Datadog/New Relic when:

You're monitoring 50+ servers
You need APM with distributed tracing
You need log aggregation and analysis
You need custom metrics and dashboards beyond server basics
Your company is paying (not your credit card)

The gap between Beszel and Datadog is features, not quality. Beszel's metrics are accurate and its dashboard is clean. It's not trying to be an enterprise observability platform - it's trying to be the best lightweight server monitor, and it succeeds.

My Setup

I run the Beszel hub on a $3/mo managed server with agents on 4 servers. Total cost: $3/mo for monitoring my entire infrastructure. The same setup on Datadog would cost $60/mo ($15/host x 4).

The dashboard gives me everything I check daily: CPU load, memory usage, disk space, network throughput, and container status. If I need deeper analysis, I SSH into the server directly.

For most independent developers and small teams, Beszel is the monitoring tool you actually need. Not the one enterprise sales teams convinced you to buy.

What monitoring tool are you using for your servers? Curious if anyone's tried switching from Datadog/Grafana to something lighter.

The Cheapest Way to Self-Host Vaultwarden in 2026

Vikas Singhal — Tue, 07 Apr 2026 09:45:58 +0000

Last updated: April 2026

Vaultwarden is the most popular self-hosted password manager in 2026 - a lightweight, Rust-based alternative to Bitwarden's official server. It's compatible with all official Bitwarden clients (browser extensions, mobile apps, desktop apps, CLI), uses a fraction of the resources, and search interest has grown 83% year-over-year. Over 42,000 GitHub stars and actively maintained.

The catch with any self-hosted password manager: it has to be online 24/7, it has to be secure, and it has to be backed up. If your Vaultwarden instance goes down, you can't log into anything. If it gets compromised, everything is compromised. The hosting choice matters more here than for almost any other self-hosted app.

I've run Vaultwarden on multiple setups. Here's every option I found, ranked by actual monthly cost, with the trade-offs that matter for a password manager specifically.

TLDR: The cheapest managed option is InstaPods at $3/mo - one-click deploy with SSL, backups, and zero maintenance. The cheapest self-managed option is a $4-5/mo Hetzner VPS with Docker. Bitwarden's official cloud starts at $0 (free tier) or $10/year for Premium. The right answer depends on whether you want to own your data and how many users you have.

Vaultwarden vs Bitwarden: Why Self-Host?

Before the hosting comparison, the obvious question: why not just use Bitwarden Cloud?

Bitwarden Cloud pricing:

Free: 1 user, unlimited passwords, basic features
Premium: $10/year per user (TOTP, file attachments, vault health reports)
Families: $40/year for 6 users
Teams: $4/user/month
Enterprise: $6/user/month

For a single person, Bitwarden's free tier is hard to beat. But self-hosting Vaultwarden makes sense when:

You have a family/team - Vaultwarden gives you unlimited users and all premium features for free. A family of 6 saves $40/year. A team of 10 saves $480/year.
You want full data control - Your vault never touches Bitwarden's servers. Useful for compliance or paranoia.
You want premium features without paying - TOTP authenticator, file attachments, emergency access, vault health reports - all free in Vaultwarden.
You're already self-hosting other apps - Adding Vaultwarden to an existing server costs nothing extra.

Every Way to Host Vaultwarden, Ranked by Cost

Method	Monthly Cost	You Manage	Setup Time	Users
Add to existing server	$0 extra	Container	~10 min	Unlimited
Oracle Cloud free tier	$0	Everything	~1 hr	Unlimited
InstaPods	$3	Nothing	~30 sec	Unlimited
PikaPods	~$2.50	Nothing	~1 min	Unlimited
Hetzner VPS + Docker	~$4-5	Everything	~45 min	Unlimited
Coolify on Hetzner VPS	~$5-8	VPS + OS	~15 min	Unlimited
Elestio	~$17	Nothing	~3 min	Unlimited
Cloudron on Hetzner VPS	~$21	VPS + OS	~40 min	Unlimited
Bitwarden Cloud (free)	$0	Nothing	~2 min	1
Bitwarden Cloud (families)	$3.33	Nothing	~2 min	6

Let me walk through each one.

1. Hetzner VPS + Docker (~$4-5/mo)

The standard self-hosted approach. Rent a VPS, install Docker, run Vaultwarden.

services:
  vaultwarden:
    image: vaultwarden/server:latest
    container_name: vaultwarden
    restart: unless-stopped
    environment:
      DOMAIN: "https://vault.yourdomain.com"
      SIGNUPS_ALLOWED: "false"
      ADMIN_TOKEN: "your-secure-admin-token-here"
    volumes:
      - vw-data:/data
    ports:
      - "8080:80"
volumes:
  vw-data:

Cost breakdown:

Hetzner CX22: EUR 3.99/mo (~$4.30) - 2 vCPU, 4 GB RAM, 40 GB storage
Or Hetzner CAX11 (ARM): EUR 3.29/mo (~$3.55) - 2 Ampere vCPU, 4 GB RAM

What you also need to set up:

Reverse proxy (Caddy recommended - automatic HTTPS)
SSL certificate (handled by Caddy, or Let's Encrypt + nginx)
Firewall rules (UFW - allow only 443 and SSH)
Automatic updates (Watchtower or cron job to pull new images)
Backups (critical - cron job to back up the /data volume to offsite storage)
Fail2ban or similar (Vaultwarden supports fail2ban log format)

Security considerations for a password manager:

Set SIGNUPS_ALLOWED=false after creating your account
Use a strong ADMIN_TOKEN (or disable the admin panel entirely with an empty value)
Enable 2FA on your Vaultwarden account immediately
Set up automated offsite backups (not just local snapshots)
Keep the VPS updated - unattended-upgrades for automatic security patches

Pros: Full control. Can run other apps on the same server. Cheapest paid option.
Cons: You're the sysadmin. A misconfigured reverse proxy or firewall exposes your password vault. The responsibility is real.

Best for: Experienced Linux admins who already manage servers and understand the security implications.

2. Oracle Cloud Free Tier ($0/mo)

Oracle offers an always-free ARM instance (4 OCPU, 24 GB RAM) that can run Vaultwarden easily. It's massively overpowered for this use case.

Cost: $0 if you can get an instance. Availability is limited - you'll likely hit "out of capacity" errors for days or weeks.

Catch: Oracle has reclaimed free-tier instances from some users without warning. Your password manager going down because Oracle decided to reclaim your VM is a worst-case scenario. Reports vary - some people have run free-tier instances for years, others lost theirs after months.

Best for: Testing Vaultwarden. Not recommended for your primary password vault unless you have reliable backups and a migration plan.

3. Coolify on Hetzner VPS (~$5-8/mo)

Install Coolify on a VPS and deploy Vaultwarden from a Docker Compose template.

Cost: VPS ~$4-8/mo depending on size. Coolify is free. Recommended: CX22 or higher since Coolify itself needs ~2 GB RAM.

Pros: Web dashboard for managing Vaultwarden. Automated SSL. Can host other apps alongside it.
Cons: Coolify adds resource overhead. You still manage the VPS itself. More moving parts = more potential failure points for a security-critical app.

Best for: People already running Coolify who want to add Vaultwarden to their stack.

4. InstaPods ($3/mo)

InstaPods has Vaultwarden as a one-click app. Click deploy, get a running instance with HTTPS in about 30 seconds. $3/mo on the Launch plan.

Full disclosure: I built InstaPods. Including it because it's genuinely one of the cheapest managed options. I'll be honest about the gaps.

Cost: $3/mo (Launch plan: 0.5 vCPU, 512 MB RAM, 5 GB storage). Vaultwarden is extremely lightweight - the $3 plan is plenty even for a family vault.

Pros: Fastest setup. SSL and custom domain included. Backups included. No server to manage. SSH access if you need to poke around.
Cons: New platform (launched 2026). Single region (EU - Nuremberg). Smaller community than established platforms.

Best for: People who want a managed Vaultwarden instance at the lowest price without touching a terminal.

5. PikaPods (~$2.50/mo)

PikaPods offers managed Vaultwarden hosting. Set your resource sliders, deploy, done.

Cost: ~$2.50/mo for Vaultwarden with default resources. PikaPods shares 30% of revenue with the Vaultwarden project.

Pros: Zero maintenance. Supports the Vaultwarden project financially. $5 welcome credit. Established platform with good track record.
Cons: Limited configuration compared to self-hosted. No SSH access.

Best for: Non-technical users who want the simplest managed path.

6. Elestio (~$17/mo)

Elestio deploys Vaultwarden on a dedicated VM with your choice of cloud provider.

Cost: ~$17/mo minimum on Hetzner. Higher on AWS, GCP, or Azure.

Pros: True zero-maintenance. Auto-updates, backups, security patches handled. Choose from 8 cloud providers.
Cons: $17/mo for a single app is steep when managed alternatives exist at $3/mo.

Best for: Teams with budget who want fully managed hosting on their preferred cloud.

7. Cloudron on Hetzner VPS (~$21/mo)

Install Cloudron on a VPS and deploy Vaultwarden from the app store.

Cost: VPS (~$5/mo) + Cloudron license (EUR 15/mo) = ~$21/mo.

Pros: Best admin dashboard of any self-hosted platform. SSO, automated backups, one-click updates.
Cons: The license fee makes this the most expensive self-hosted option. Only worth it if you're running 5-10 apps on the same Cloudron server.

Best for: Existing Cloudron users. Not worth the license for Vaultwarden alone.

8. Adding Vaultwarden to an Existing Server ($0 extra)

Already running a VPS or self-hosted platform for other apps? Adding Vaultwarden costs nothing extra. It's one of the lightest self-hosted apps out there - 50 MB RAM, minimal CPU.

Add a Vaultwarden service to your existing Docker Compose stack, point a subdomain at it, and you're done. This is how most people end up running Vaultwarden - it piggybacks on infrastructure they already have.

Best for: Anyone who already has a server running other self-hosted apps.

The Real Comparison

Here's what it comes down to:

If you just need a personal password manager:
Bitwarden Cloud's free tier is hard to beat. You get unlimited passwords, sync across all devices, and zero maintenance. The $10/year Premium plan adds TOTP and file attachments.

If you have a family or small team:
Self-hosted Vaultwarden saves money fast. PikaPods (~$2.50/mo) and InstaPods ($3/mo) are the cheapest managed options. Unlimited users, all premium features, your data on a server you control. A family of 6 on Vaultwarden ($3/mo = $36/year) saves $4/year vs Bitwarden Families ($40/year) - modest. But a team of 10 saves $480/year vs Bitwarden Teams ($4/user/month = $480/year), and the gap grows with every user you add.

If you want full control:
Hetzner VPS + Docker at ~$4-5/mo. You manage everything, including security - which matters more for a password manager than for most apps.

If security is your top priority:
Honestly? Bitwarden Cloud. Their infrastructure is audited, they have a dedicated security team, and they've been doing this for years. Self-hosting means you are the security team. If you're confident in your ability to keep a server patched, a firewall configured, and backups verified, self-host. If not, Bitwarden Cloud is the safer choice.

Vaultwarden Security Checklist

If you self-host, run through this list before sharing access with anyone:

[ ] SIGNUPS_ALLOWED=false after creating your account(s)
[ ] Strong ADMIN_TOKEN set (or admin panel disabled)
[ ] 2FA enabled on all Vaultwarden accounts
[ ] HTTPS enforced (never run a password manager over HTTP)
[ ] Firewall configured (only ports 443 and SSH open)
[ ] Automated offsite backups running and tested
[ ] OS auto-updates enabled (unattended-upgrades on Ubuntu)
[ ] Vaultwarden image auto-updates configured (Watchtower or cron)
[ ] Fail2ban configured for brute-force protection
[ ] Emergency access plan (what happens if the server dies and you can't log into anything?)

That last point is critical. Export your vault regularly and store the encrypted export somewhere safe that doesn't require Vaultwarden to access.

FAQ

How much RAM does Vaultwarden need?
Almost nothing. Vaultwarden uses about 50 MB of RAM in normal operation. Even the smallest VPS or managed plan runs it comfortably. This is one of the lightest self-hosted apps you can run - the $3/mo tier on any managed platform is more than enough.

Is Vaultwarden as secure as Bitwarden?
Vaultwarden implements the same client-server protocol as Bitwarden. Your vault is encrypted client-side before it ever reaches the server - even the server operator (you) can't read your passwords. The difference is infrastructure security: Bitwarden Cloud has a dedicated security team and regular audits. Self-hosted security depends entirely on you.

Can I use official Bitwarden apps with Vaultwarden?
Yes. All official Bitwarden clients work with Vaultwarden - browser extensions (Chrome, Firefox, Safari, Edge), mobile apps (iOS, Android), desktop apps (Windows, Mac, Linux), and the CLI. Just point the client to your Vaultwarden URL instead of vault.bitwarden.com.

How do I migrate from Bitwarden Cloud to Vaultwarden?
Export your vault from Bitwarden Cloud (Settings > Export Vault), then import the JSON file into Vaultwarden. All passwords, notes, cards, and identities transfer over. Attachments need to be re-uploaded manually.

How do I back up Vaultwarden?
Back up the /data directory (Docker volume). It contains the SQLite database with all vault data, attachments, and configuration. Use a cron job to copy this to offsite storage (S3, Backblaze B2, or another server) daily. On managed platforms like PikaPods and InstaPods, backups are handled automatically.

What happens if my Vaultwarden server goes down?
Bitwarden clients cache your vault locally. You can still access passwords offline on any device that has synced recently. But you can't sync new passwords, share items, or log in on a new device until the server is back. This is why uptime matters more for a password manager than most apps.

Should I use Vaultwarden or the official Bitwarden self-hosted server?
Vaultwarden if you want minimal resource usage, easy Docker setup, and all premium features for free. Official Bitwarden self-hosted server if you need enterprise features (SSO/SCIM, event logs, policy management) or want official support. The official server needs 4+ GB RAM and Docker Compose with multiple containers - it's heavier to run.

For a broader comparison of self-hosted platforms (not just for Vaultwarden), I wrote about Coolify vs Cloudron vs CapRover and a 6-platform comparison.

If you want to try the managed route, InstaPods has Vaultwarden as a one-click app - $3/mo, HTTPS included, deployed in 30 seconds.

What's your Vaultwarden setup? Running it on its own server or piggybacking on an existing stack? Drop it in the comments.

n8n vs Zapier in 2026: I Switched and Cut My Automation Bill by 95%

Vikas Singhal — Fri, 20 Mar 2026 07:51:32 +0000

Last updated: March 2026

I was paying Zapier $69/mo to run about 15 workflows. Most of them were simple - sync a form submission to a spreadsheet, post a Slack message when a GitHub issue gets labeled, send a weekly digest from an RSS feed. Nothing fancy.

Then I tried n8n. Same workflows, same results. My bill went from $69/mo to $3/mo.

Here's the full comparison so you can decide if the switch makes sense for you.

TLDR: n8n is an open-source workflow automation tool that does 90% of what Zapier does. Zapier wins on integrations (7,000+ vs 400+ built-in) and ease of setup (zero learning curve). n8n wins on pricing (free to self-host vs $20-69/mo), features (code nodes, conditional logic on all plans), and data ownership. If you're comfortable self-hosting or using a managed platform, n8n saves you hundreds to thousands per year.

The Quick Comparison

	Zapier	n8n
Starting price	$20/mo (Starter)	Free (self-hosted)
Task/execution limits	750 tasks/mo on Starter	Unlimited
Workflows	Unlimited (Starter+)	Unlimited
Multi-step workflows	Starter plan and above	All plans
Conditional logic (paths)	Professional plan ($49/mo)	Included on all plans
Code nodes	JavaScript only (paid plans)	JavaScript + Python
Integrations	7,000+ apps	400+ built-in, 900+ community
Self-hosting	Not available	Full control
Data retention	Limited by plan tier	Unlimited (your server)
Source code	Closed	Open source (fair-code)
Webhooks	All paid plans	Included
Error handling	Professional+	All plans

Pricing: Where the Real Difference Lives

This is the biggest reason I switched. Zapier's pricing gets expensive fast, and the per-task model means your bill grows with your usage.

Zapier's pricing tiers (March 2026):

Free: 100 tasks/mo, 5 single-step Zaps
Starter: $20/mo for 750 tasks
Professional: $49/mo for 2,000 tasks (adds paths, filters, custom logic)
Team: $69/mo for 2,000 tasks (adds shared workspaces, SSO)

The catch with Zapier: a workflow with 5 steps counts as 5 tasks per run. So a "simple" workflow that runs 10 times a day burns through 50 tasks. At that rate, the 750-task Starter plan lasts about 15 days before you need to upgrade.

n8n's pricing options:

Self-hosted: Free forever. You provide the server.
n8n Cloud: Starts at $24/mo (2,500 executions). Their hosted version with support.
Self-hosted on InstaPods: $3/mo. One-click deploy, they handle the server.
Self-hosted on PikaPods: ~$3.80/mo. Similar managed hosting.

My actual numbers: I was on Zapier Professional at $69/mo ($828/year). Now I run the same workflows on a self-hosted n8n instance for $3/mo ($36/year). That's a 95% cost reduction.

And there are no task limits. My workflows run as often as they need to. Some fire every 5 minutes, some run on webhooks dozens of times a day. No meter ticking.

Features: What You Get on Each Platform

Where Zapier is better

More native integrations. Zapier has 7,000+ app connections. If an app has an API, Zapier probably has a pre-built integration for it. n8n has 400+ built-in nodes plus 900+ community-contributed ones. For mainstream tools (Slack, Google Sheets, GitHub, Notion, Airtable), both work fine. For niche apps, Zapier has better coverage.

Easier initial setup. Zapier requires zero technical knowledge. Pick a trigger, pick an action, connect your accounts, done. n8n's visual editor is intuitive, but there's a slight learning curve - maybe 30 minutes to feel comfortable.

No hosting to think about. Zapier is a pure SaaS product. Nothing to install, nothing to maintain, nothing to update. It works.

Where n8n is better

No per-task billing. This is the big one. Run 100 workflows or 100,000 executions - your cost stays the same on self-hosted n8n.

Code nodes on every plan. n8n lets you write JavaScript or Python in any workflow, on any plan. Zapier limits code steps to paid plans and only supports JavaScript.

Conditional logic included. Zapier locks paths (conditional branching) behind the $49/mo Professional plan. n8n includes IF/Switch nodes on every plan, including self-hosted.

Full data ownership. Your workflow data, execution logs, and API credentials stay on your server. Nothing leaves your infrastructure.

Error handling everywhere. n8n has error workflows, retry logic, and manual execution replay on all plans. Zapier gates advanced error handling behind higher tiers.

AI-ready. n8n has built-in nodes for OpenAI, Anthropic, and other AI providers. Build AI agents and chains inside your workflows. Zapier has AI features too, but they're newer and more limited.

Self-Hosting: The Make-or-Break Question

Here's where the decision gets personal. Self-hosting n8n means you need somewhere to run it. You have three paths:

Option 1: Manage your own server

Rent a VPS from Hetzner (~$4-5/mo), install Docker, run n8n with docker-compose. You handle updates, backups, SSL, and monitoring.

Good for: Developers comfortable with Linux who want full control.
Time cost: ~45 minutes to set up, plus ongoing maintenance.

Option 2: Use a managed hosting platform

Platforms like InstaPods ($3/mo) and PikaPods (~$3.80/mo) let you deploy n8n with one click. They handle the server, SSL, and updates. You get n8n running in under a minute.

Full disclosure: I built InstaPods. Including it because it's the cheapest managed n8n hosting I've found, but PikaPods is a solid alternative with a longer track record.

Good for: Anyone who wants self-hosted n8n without the ops work.
Time cost: Under a minute.

Option 3: Use n8n Cloud

n8n's official hosted version starts at $24/mo. More expensive than self-hosting but includes official support and collaboration features. Still cheaper than Zapier Professional.

Good for: Teams that need official n8n support or multi-user collaboration.

I wrote a deeper breakdown of every way to host n8n ranked by cost if you want the full picture.

Integrations: Quality vs Quantity

Zapier's 7,000+ integrations vs n8n's 400+ built-in sounds like a blowout. It's more nuanced than that.

For common tools, both work equally well. Slack, Google Workspace, GitHub, Notion, Airtable, Stripe, HubSpot, Salesforce, Jira - all covered on both platforms.

For niche tools, Zapier wins. If you need a pre-built integration with an obscure SaaS product, Zapier is more likely to have it.

n8n's escape hatch is strong. The HTTP Request node lets you connect to any REST API. If a tool has an API (and most do), you can integrate it in n8n - it takes a few more minutes than a pre-built integration but works for anything.

Community nodes fill gaps. n8n's 900+ community nodes cover many tools that aren't in the core set. Installing community nodes is a one-click process on self-hosted instances.

Bottom line: If your workflows use mainstream tools, n8n's integration coverage is more than enough. If you rely on niche SaaS products with no public API, check n8n's node list before switching.

Ease of Use

Both platforms have visual workflow builders. You drag nodes onto a canvas, connect them, and configure each step.

Zapier's approach: Linear, step-by-step. Great for simple automations. The interface hides complexity, which makes it approachable but limiting for advanced workflows.

n8n's approach: Freeform canvas. You can branch, loop, merge, and build complex data pipelines. More powerful, slightly steeper learning curve.

My honest take: If you've never built an automation before, Zapier will feel easier for the first hour. If you've built a few Zaps and hit their limits (no branching without upgrading, can't debug intermediate steps easily), n8n will feel like freedom.

n8n's workflow execution view is something Zapier doesn't match. You can see the exact data flowing through each node, re-run individual steps, and debug problems by inspecting real payloads. On Zapier, debugging means re-running the entire Zap and hoping the logs show enough detail.

Migration: Moving from Zapier to n8n

Switching isn't instant, but it's not painful either. There's no automatic import tool - you rebuild workflows in n8n's editor.

What to expect:

Simple Zaps (trigger + action): 5-10 minutes each to rebuild in n8n
Multi-step Zaps with filters: 15-30 minutes each
Complex Zaps with paths and code: 30-60 minutes each

Tips from my migration:

Start with your most expensive Zaps (the ones burning the most tasks)
Use n8n's template library - many common workflows are pre-built
Re-enter API credentials manually (they don't export from Zapier for security)
Run both platforms in parallel for a week before cutting over

I migrated 15 workflows in about 4 hours total. Given the $800/year savings, the ROI was clear within the first month.

When to Stick with Zapier

I don't think n8n is the right choice for everyone. Stick with Zapier if:

You're non-technical and need it to work. Zapier's learning curve is near zero. n8n's is 30 minutes to an hour.
You depend on niche integrations. If your workflows rely on 5+ apps that only Zapier supports, rebuilding with HTTP nodes in n8n adds friction.
Your team needs zero maintenance. Zapier is pure SaaS. Even managed n8n hosting requires picking a provider and an occasional update check.
You're under 100 tasks/month. Zapier's free tier handles light usage fine. No reason to switch if it's working.

When n8n Is the Clear Winner

Switch to n8n if:

Your Zapier bill is over $20/mo. At that point, self-hosted n8n saves real money.
You hit task limits regularly. n8n has no per-task fees. Run unlimited executions.
You need code in your workflows. n8n's JavaScript + Python code nodes are available on every plan. Zapier locks them behind paid tiers.
Data privacy matters. Self-hosted n8n keeps everything on your infrastructure. Your API keys and workflow data never touch a third party.
You want conditional logic without paying $49/mo. Branching, filtering, and error handling are included on all n8n plans.

FAQ

Is n8n hard to set up?

It depends on your path. On a managed platform like InstaPods or PikaPods, setup takes under a minute - click deploy, get a running instance with HTTPS. On a raw VPS with Docker, budget 30-45 minutes for setup (Docker, reverse proxy, SSL). n8n Cloud takes about 2 minutes.

Can n8n replace Zapier completely?

For most users, yes. n8n covers the same core functionality - triggers, actions, multi-step workflows, webhooks, scheduling. The gap is in niche integrations. If your critical workflows use tools that only Zapier supports, you might need the HTTP Request node as a workaround or keep a lightweight Zapier plan alongside n8n.

Is n8n free?

n8n is open source under a "fair-code" license (Sustainable Use License). You can self-host it for free on your own server. n8n Cloud (their hosted version) starts at $24/mo. Managed hosting on platforms like InstaPods runs $3/mo.

How does n8n handle errors compared to Zapier?

n8n has built-in error workflows, automatic retries with configurable delays, and a manual execution replay feature. You can inspect the exact data at each step when something fails. Zapier has error notifications and auto-replay on some plans, but the debugging experience is more limited.

What happens to my workflows if n8n goes away?

Your workflows are stored as JSON files on your server. You own them completely. Export, back up, and version control them like any other code. This is the opposite of Zapier, where workflows can't be exported to another platform.

If you're exploring self-hosting, these might help:

The Cheapest Way to Self-Host n8n in 2026 - every hosting option ranked by cost
I Compared 6 Platforms for Deploying Self-Hosted Apps in 2026 - broader self-hosted platform comparison
Coolify vs Cloudron vs CapRover in 2026 - if you're considering self-hosted PaaS for n8n and other apps

What automation platform are you using? Have you tried switching from Zapier to n8n (or the other way)? I'd love to hear how it went in the comments.

AI Can Build Your App in 5 Minutes. Why Does Deploying It Still Take 2 Hours?

Vikas Singhal — Thu, 19 Mar 2026 19:15:27 +0000

Last updated: March 2026

Something shifted last year.

Tools like Claude Code, Cursor, and Lovable made it possible to go from idea to working app in minutes. Not a wireframe. Not a mockup. A real, running application with a backend, database calls, and authentication.

I built a full recipe app with Claude Code in under 10 minutes. Frontend, backend, database - the whole thing.

Then came the question everyone asks next:

"Cool. How do I get this online?"

And that's where the magic dies.

The Ratio Is Broken

Think about this:

Building the app: 5-15 minutes with AI
Deploying the app: 30 minutes to 2 hours manually

The build phase got 100x faster. The deploy phase hasn't changed since 2015. You still need to:

Provision a server
Install your runtime (Node, Python, PHP)
Configure a reverse proxy
Set up SSL certificates
Configure DNS
Set up a process manager
Figure out how to get your files there

Seven steps. For an app that took 5 minutes to build.

"Just Use Vercel"

I hear this a lot. And Vercel is great - if you're deploying a Next.js frontend.

But AI tools don't generate Vercel-shaped apps. They generate full-stack apps. A Python Flask API with SQLite. A Node.js server with file uploads. A PHP app with a MySQL database.

Try deploying a Flask app with a SQLite database to Vercel. You can't. The model doesn't fit.

"Just Use a VPS"

Sure. Spin up a DigitalOcean droplet. SSH in. Install Node. Configure nginx. Set up certbot. Create a systemd service. Configure your firewall. Set up a domain. That's 7 steps and 45 minutes for an app that took 5 minutes to build.

The person who just built an app by talking to an AI doesn't want to become a sysadmin. That's the whole point - they used AI because they wanted to skip the tedious parts.

"Just Use Railway/Render/Fly"

Closer. But these platforms still expect you to understand Dockerfiles, build configs, and environment variables. They're simpler than a raw VPS, but they're built for developers who already know DevOps basics.

And the pricing. Railway and Fly.io use usage-based billing - you don't know what you'll pay until the bill arrives. One person on Reddit reported a $45 Railway bill for a side project that went viral for a day. Not great for side projects.

What Deploy Should Look Like in 2026

If AI compressed building to minutes, deploying should match. Here's what I think that looks like:

One command. Not a 12-step guide. Not a Dockerfile. One command that detects your stack and handles everything.

Real servers. AI tools generate apps that need filesystems, databases, and background processes. You need an actual Linux environment, not a sandboxed function.

Zero config. No YAML. No build pipelines. No environment variable dashboards. Push code, it runs.

Agent-native. If an AI built the app, an AI should be able to deploy it. CLI tools and APIs that agents can call directly - not web dashboards that require clicking through a UI.

Flat pricing. $3/mo means $3/mo. People building side projects with AI tools need predictable costs, not usage-based billing that scales with traffic.

What I'm Building

This problem bugged me enough that I built a solution. InstaPods is a hosting platform where you pick a stack (Node, PHP, Python, static), get a real Linux server, and your code is live.

$ instapods deploy my-app

  Deploying my-app
  Detected nodejs (package.json)

  Creating pod ·················· ✓ 1.2s
  42 files uploaded ············· ✓ 0.8s
  Reloading ····················· ✓ 1.4s

  ✓ Deployed in 3.4s
  → https://my-app.instapods.app

One command. No configuration. Pods launch instantly. $3/mo flat - no usage billing, no bandwidth charges. $10 free credit on signup, no credit card required.

Every pod is a real Linux server with SSH access. Not a sandboxed function, not a locked-down platform. You can SSH in, install packages, run databases - whatever you need.

It also has an MCP server, so AI agents like Claude Code and Cursor can create pods and deploy code directly from the conversation without you touching a terminal.

I'm building this solo, and it's still early. But the core experience works: go from code to live URL in seconds, not hours.

The Bigger Picture

AI made building 100x faster, but the infrastructure around it hasn't caught up. Deploying, monitoring, scaling - all still manual.

The entire deploy experience is going to get rebuilt over the next couple of years. Not by adding AI features to existing platforms, but by building new platforms that assume AI is doing the building.

FAQ

Can AI coding tools deploy apps directly?
Some can. Claude Code and Cursor can deploy through MCP (Model Context Protocol) - they connect to hosting platforms and deploy code directly from the conversation. InstaPods has an MCP server with 8 tools for creating pods, pushing code, and managing apps without leaving your editor.

What kind of hosting do vibe-coded apps need?
Most AI-generated apps are full-stack - Express.js or Flask backends with SQLite or PostgreSQL databases. They need a host with a persistent filesystem and the ability to run a long-lived process. Serverless platforms (Vercel, Netlify) won't work. You need either a VPS, a PaaS like Railway, or a managed platform like InstaPods.

Why can't I deploy my AI-generated app to Vercel?
Vercel runs Node.js as stateless serverless functions. There's no persistent filesystem, so SQLite databases get wiped on every cold start. File uploads disappear. Background jobs can't run. If your AI tool generated a standard Express/Flask/Django app, you need a different kind of host.

For hands-on comparisons, I tested 5 platforms deploying the same vibe-coded app, compared Coolify vs Cloudron vs CapRover, and wrote about the cheapest way to self-host n8n.

What's your deploy workflow for AI-generated apps? Drop a comment - I'm curious what's working for people.

The Cheapest Way to Self-Host Uptime Kuma in 2026

Vikas Singhal — Thu, 19 Mar 2026 19:06:52 +0000

Last updated: March 2026

Uptime Kuma is the best open source uptime monitoring tool you can self-host. It's a self-hosted alternative to UptimeRobot, Pingdom, and StatusCake - unlimited monitors, beautiful status pages, and notifications to 70+ services including Slack, Discord, Telegram, and email. Over 65,000 GitHub stars and actively maintained.

But "self-hosted" means you need somewhere to run it. And monitoring tools have a requirement most apps don't - they need to be running 24/7 with zero downtime. Your uptime monitor going down is the one outage nobody notices until everything else is also down.

I've tried multiple hosting setups for Uptime Kuma. Here's every option I found, ranked by actual monthly cost, with the trade-offs that matter for a monitoring tool specifically.

TLDR: The cheapest managed option is InstaPods at $3/mo or PikaPods at ~$2.50/mo - both handle everything. The cheapest self-managed option is a $3-5/mo Hetzner VPS with Docker. UptimeRobot's paid plan starts at $7/mo. The right answer depends on how many monitors you need and whether you want to own your data.

Every Way to Host Uptime Kuma, Ranked by Cost

Method	Monthly Cost	You Manage	Setup Time	Monitors
Oracle Cloud free tier	$0	Everything	~1 hr	Unlimited
PikaPods	~$2.50	Nothing	~1 min	Unlimited
InstaPods	$3	Nothing	~30 sec	Unlimited
Hetzner VPS + Docker	~$3.50-5	Everything	~30 min	Unlimited
Coolify on Hetzner VPS	~$5-8	VPS + OS	~10 min	Unlimited
CapRover on Hetzner VPS	~$5-6	Everything	~20 min	Unlimited
UptimeRobot (paid)	$7+	Nothing	~2 min	50+
Render	~$7	Nothing	~10 min	Unlimited
Elestio	~$17	Nothing	~3 min	Unlimited
Cloudron on Hetzner VPS	~$21	VPS + OS	~30 min	Unlimited

Let me walk through each one.

1. Hetzner VPS + Docker (~$3.50-5/mo)

The DIY approach. Rent a cheap VPS, install Docker, run Uptime Kuma.

version: "3"
services:
  uptime-kuma:
    image: louislam/uptime-kuma:1
    ports:
      - "3001:3001"
    volumes:
      - uptime-kuma-data:/app/data
    restart: always
volumes:
  uptime-kuma-data:

Cost breakdown:

Hetzner CX11: EUR 3.29/mo (~$3.55) - 1 vCPU, 2 GB RAM, 20 GB storage
Or Hetzner CAX11 (ARM): EUR 3.29/mo (~$3.55) - 2 Ampere vCPU, 4 GB RAM

Uptime Kuma is lightweight. It runs comfortably on the smallest VPS. Unlike n8n which needs at least 512 MB, Uptime Kuma runs fine with 256 MB of RAM even with 50+ monitors.

What you also need to set up:

Reverse proxy (nginx or Caddy) for HTTPS
SSL certificate (Let's Encrypt + certbot)
Firewall rules (UFW)
Automatic updates for Uptime Kuma and the OS
Backups (cron job for the /app/data directory)
Monitoring for your monitoring tool (yes, really - who watches the watchers?)

Pros: Cheapest paid option. Full control. Extremely lightweight. Can run other services on the same VPS.
Cons: You're the sysadmin. If the VPS has an outage, your monitoring goes down and you won't know your other services are down either.

Best for: Developers comfortable with Linux and Docker who want full control and don't mind maintaining the server.

2. Oracle Cloud Free Tier ($0/mo)

Oracle offers an always-free ARM instance (4 OCPU, 24 GB RAM) that can run Uptime Kuma. It's genuinely free - not a trial.

Cost: $0 if you can get an instance. Availability is limited and you'll hit "out of capacity" errors for days or weeks before getting one.

Catch: Oracle's free tier has been reliable for some people and randomly terminated for others. There are reports of instances being reclaimed without warning. Running your uptime monitoring on infrastructure that might disappear is... not ideal.

The deeper problem: Your monitoring tool needs to be more reliable than the things it monitors. Putting it on the least reliable infrastructure available defeats the purpose.

Best for: Testing Uptime Kuma or monitoring non-critical personal projects. Not recommended for anything you depend on.

3. Coolify on Hetzner VPS (~$5-8/mo)

Install Coolify (free, open source) on a VPS and deploy Uptime Kuma through its one-click catalog.

Cost breakdown:

Hetzner CX22: ~$4.30/mo (minimum for Coolify + Uptime Kuma on same server)
Hetzner CX32: ~$7.50/mo (recommended - Coolify itself uses 2 GB RAM)

Pros: Web UI for managing Uptime Kuma. Auto-updates. Can run other apps alongside it. Free.
Cons: Coolify's 2 GB RAM overhead is overkill when Uptime Kuma itself needs 256 MB. You're running a management platform heavier than the app it manages.

Best for: People already running Coolify for other apps who want to add monitoring.

4. CapRover on Hetzner VPS (~$5-6/mo)

Install CapRover (free) and deploy Uptime Kuma from its template catalog.

Cost: Same as DIY Docker - just the VPS cost. CapRover is free.

Pros: One-click Uptime Kuma install. Web UI.
Cons: CapRover development has stalled (last commit Dec 2025). Docker Swarm dependency is a long-term risk. You still manage the VPS.

Best for: Existing CapRover users. For new setups, pick Coolify instead.

5. InstaPods ($3/mo)

InstaPods has Uptime Kuma as a one-click app. Click deploy, get a running instance with SSL in about 30 seconds. $3/mo on the Launch plan.

Full disclosure: I built InstaPods. Including it because it's genuinely one of the cheapest managed options and the comparison wouldn't be complete without it. I'll be honest about the limitations.

Cost: $3/mo (Launch plan: 0.5 vCPU, 512 MB RAM, 5 GB storage). More than enough for Uptime Kuma - it barely uses 150 MB of RAM even with dozens of monitors.

Pros: Fastest setup of any option. No server management. SSL and URL included. Uptime Kuma runs on a real server (not a serverless function that sleeps). $10 free credit on signup.
Cons: New platform (launched 2026). Single region (EU - Nuremberg). Smaller community than established platforms.

Best for: People who want Uptime Kuma running in under a minute with zero maintenance.

6. PikaPods (~$2.50/mo)

PikaPods is managed hosting for open source apps. Pick Uptime Kuma, set your resource sliders, done.

Cost: ~$2.50/mo for Uptime Kuma with default resources. Uptime Kuma is lightweight enough that the minimum resources work fine.

Pros: Zero maintenance. Revenue sharing with open source projects (part of your payment goes directly to Uptime Kuma's developer). $5 welcome credit. Established platform with a good track record.
Cons: Limited configuration options. Can't customize Uptime Kuma's environment beyond what PikaPods exposes.

Best for: Non-technical users who want managed Uptime Kuma at the lowest price. Probably the easiest option overall.

7. UptimeRobot Paid Plan ($7+/mo)

Wait - this isn't self-hosting. True. But most people looking at Uptime Kuma are comparing it to UptimeRobot, so the pricing context matters.

UptimeRobot free tier: 50 monitors, 5-minute check intervals. No status page customization. Limited notification options.

UptimeRobot Pro: $7/mo for 50 monitors with 1-minute intervals. More notification methods. Custom status pages.

Why people switch to Uptime Kuma:

UptimeRobot keeps reducing free tier value - check intervals have gotten slower over the years, and useful features keep moving behind the paywall
UptimeRobot Pro at $7/mo adds up when self-hosted Uptime Kuma gives you unlimited everything for $2.50-3/mo
Uptime Kuma has 70+ notification providers vs UptimeRobot's ~12
Data privacy - your monitoring data stays on your server
No vendor lock-in

The math: UptimeRobot Pro costs $7/mo for 50 monitors with 1-minute intervals. Self-hosted Uptime Kuma on PikaPods ($2.50/mo) or InstaPods ($3/mo) gives you unlimited monitors with 1-second check intervals - more features for less than half the price.

8. Render (~$7/mo)

Render can run Uptime Kuma as a Docker web service.

Cost: Free tier won't work - Render's free services sleep after 15 minutes of inactivity, and a monitoring tool that sleeps defeats the purpose. The paid Individual plan starts at $7/mo.

Important: You also need a Render Disk ($0.25/GB/mo) or your Uptime Kuma data resets on every deploy.

Pros: Managed. Good developer UX. Auto-deploys from Docker image.
Cons: $7/mo is expensive for a monitoring tool when managed alternatives exist at $2.50-3/mo. The sleeping free tier is useless for monitoring.

Best for: Developers already using Render for other services who want to add monitoring to their stack.

9. Elestio (~$17/mo)

Elestio is fully managed hosting. Pick Uptime Kuma, pick a cloud provider, they deploy it on a dedicated VM.

Cost: ~$17/mo minimum (Hetzner-backed VM). Higher for other cloud providers.

Pros: True zero-maintenance. Auto-updates, backups, security patches. Choose from 8 cloud providers.
Cons: $17/mo for a monitoring tool is hard to justify when the managed alternatives are $2.50-3/mo. One app = one VM is wasteful for something this lightweight.

Best for: Teams with budget who want fully managed hosting on a specific cloud provider. Otherwise overpriced for Uptime Kuma.

10. Cloudron on Hetzner VPS (~$21/mo)

Install Cloudron on a VPS and deploy Uptime Kuma from the app store.

Cost: VPS (~$5/mo) + Cloudron license (EUR 15/mo) = ~$21/mo.

Pros: Best admin dashboard. SSO, email, backups built in.
Cons: EUR 15/mo license for running a monitoring tool that needs 256 MB of RAM is absurd. Only makes sense if you're already running 5-10 other apps on Cloudron.

Best for: Existing Cloudron users only. Never worth the license just for Uptime Kuma.

The Real Comparison

Here's what it comes down to:

If you want the absolute cheapest managed option:
PikaPods at ~$2.50/mo or InstaPods at $3/mo. Both give you a running Uptime Kuma instance with SSL, zero maintenance, unlimited monitors. PikaPods has the slight price edge, InstaPods has faster deploy time.

If you want full control and have Linux skills:
Hetzner VPS + Docker at ~$3.50/mo. You manage everything, but you own every layer. Run other apps on the same server to share the cost.

If you're comparing to UptimeRobot:
Self-hosted Uptime Kuma on PikaPods or InstaPods costs less than half of UptimeRobot Pro, gives you unlimited monitors, 1-second check intervals, and 70+ notification integrations. The only thing UptimeRobot has that self-hosted doesn't is monitoring from multiple geographic locations (though you can work around this with multiple Uptime Kuma instances).

Special consideration for monitoring tools:
Your uptime monitor should be running on different infrastructure than the services it monitors. If your VPS hosts both your app AND Uptime Kuma, a server outage takes out both. You won't get notified. Consider running Uptime Kuma on a separate $3/mo managed platform even if you self-host everything else.

FAQ

How much RAM does Uptime Kuma need?
Very little. 150-200 MB for typical usage (50-100 monitors). Even with 200+ monitors, it rarely exceeds 300 MB. It's one of the lightest self-hosted apps you can run.

Can I monitor sites from multiple locations?
Not with a single instance. Uptime Kuma monitors from wherever it's running. For multi-location monitoring, you'd need multiple instances or use a service like UptimeRobot. For most people, single-location monitoring is fine.

How is Uptime Kuma different from UptimeRobot?
Uptime Kuma is self-hosted and free. UptimeRobot is a SaaS. Uptime Kuma gives you unlimited monitors, 1-second check intervals, and 70+ notification integrations. UptimeRobot's free tier gives you 50 monitors with 5-minute intervals. Uptime Kuma's status pages are more customizable.

Does Uptime Kuma support HTTPS monitoring?
Yes. It supports HTTP(S), TCP, Ping, DNS, Docker, Steam Game Server, MQTT, and more monitor types. It verifies SSL certificates and can alert you before they expire.

Can I set up a public status page?
Yes. Uptime Kuma has built-in status pages you can make public. Add your monitors, customize the look, and share the URL. No extra setup needed.

How do I back up Uptime Kuma?
Uptime Kuma stores everything in a SQLite database at /app/data/kuma.db. Back up that file. On Docker, back up the mounted volume. On managed platforms (PikaPods, InstaPods, Elestio), backups are handled for you.

Can I migrate from UptimeRobot to Uptime Kuma?
There's no official import tool, but Uptime Kuma's API lets you create monitors programmatically. Community scripts exist that export from UptimeRobot and import into Uptime Kuma. Notifications need to be set up manually since credentials don't transfer.

Should I run Uptime Kuma on the same server as my apps?
Ideally not. If that server goes down, your monitoring goes down with it and you won't get alerted. Running Uptime Kuma on a separate $3/mo managed platform is cheap insurance.

If you're looking at alternatives to traditional hosting platforms, I wrote a Heroku alternatives comparison on our blog. For more self-hosted platform comparisons, check out Coolify vs Cloudron vs CapRover and a 6-platform comparison. If you're also running n8n, check out the cheapest way to self-host n8n.

What's your monitoring setup? Running Uptime Kuma or something else? Drop it in the comments.

I Built an App With AI in 10 Minutes. Then Tried to Deploy It 5 Different Ways.

Vikas Singhal — Thu, 19 Mar 2026 19:06:44 +0000

Last updated: March 2026

I built a full-stack expense tracker with Claude Code in about 10 minutes. Express.js backend, SQLite database, vanilla HTML/CSS/JS frontend. Around 400 lines of code. It worked perfectly on localhost.

Then I spent the next two days trying to get it online.

I deployed the same app on 5 different platforms and timed everything - from signup to a working public URL. Every error, every workaround, every moment where the platform fought against what my AI-generated app needed.

TLDR: Vercel couldn't run it at all (no persistent filesystem for SQLite). Railway worked but usage-based billing made costs unpredictable. Render worked but the free tier sleeps and paid starts at $14/mo with a database. A raw DigitalOcean droplet took 47 minutes of nginx/SSL/systemd setup. InstaPods (full disclosure: mine) deployed in one command for $3/mo. The deploy gap is still the hardest part of vibe coding.

What I Built

I told Claude Code: "Build me a simple expense tracker. Express.js backend, SQLite for storage, vanilla HTML frontend. Users can add expenses with a category, amount, and date. Show a list and a total."

10 minutes later I had:

server.js - Express API with CRUD routes
expenses.db - SQLite database (created on first run)
public/index.html - Simple form + table UI
package.json - express and better-sqlite3 as dependencies

Nothing fancy. No auth, no React, no build step. A working app that reads and writes to a file-based database.

This is the kind of app people build every day with Cursor, Claude Code, and Lovable - and the kind that breaks the moment you try to deploy it anywhere serverless.

The Results

Platform	Worked?	Time to Deploy	Monthly Cost	What Went Wrong
Vercel	No	-	-	No persistent filesystem
Railway	Yes	~8 min	$5-15 (usage)	Nothing, but billing unpredictable
Render	Yes	~12 min	$7 + $7/mo	Free tier sleeps, need paid DB
DigitalOcean	Yes	~47 min	~$6/mo	Manual nginx, SSL, systemd setup
InstaPods	Yes	~2 min	$3/mo	New platform, single region

Now let me walk through each one.

1. Vercel — "Just Use Vercel"

Every time someone asks "how do I deploy this?" on Reddit, someone replies "just use Vercel." So I tried Vercel first.

What happened:

Vercel wants a framework - Next.js, Nuxt, SvelteKit, or at minimum a static export. My app is a plain Express.js server. Vercel can technically run Node.js using serverless functions, but that means restructuring my server.js into an api/ directory with individual function files.

I did that. Got the API routes working as serverless functions.

Then hit the real problem: SQLite doesn't work on Vercel. Vercel functions are stateless - every invocation gets a fresh filesystem. My SQLite database file gets created, written to, and then deleted when the function cold-starts. Every expense I add disappears within minutes.

The fix? Replace SQLite with Vercel Postgres or Vercel KV. But that means rewriting my database layer, and Vercel Postgres costs $20/mo after the hobby tier.

Result: Abandoned. My 10-minute app would need a full database rewrite to fit Vercel's model. The whole point of vibe coding is "it works" - not "it works after you swap SQLite for a managed Postgres service."

Time spent: ~25 minutes before giving up.

2. Railway — It Actually Works

Railway is a PaaS that runs your app on actual compute. Push code, it builds and deploys. No serverless restrictions.

What happened:

Connected my GitHub repo. Railway detected package.json, ran npm install, started the server. My Express app booted. SQLite worked because Railway gives you a persistent filesystem.

Time to working URL: ~8 minutes (including signup).

The catch: billing.

Railway charges by compute time, memory usage, and egress. My idle expense tracker costs maybe $1-2/mo. But Railway's minimum is $5/mo (Pro plan), and if the app gets any real traffic, costs spike. One person on Reddit reported a $45 bill for a side project that went viral for a day.

You can set spending caps, but you have to remember to do it. And there's no SSH access to debug when things go wrong.

Monthly cost: $5/mo minimum, potentially much more.

Verdict: Works well. Billing model is the only real problem. If you're OK with usage-based pricing, Railway is solid.

3. Render — Works, But Nickel-and-Dimes You

Render is another PaaS, similar to Railway but with a free tier.

What happened:

Connected GitHub, Render detected Node.js, built and deployed. App started. But my SQLite database reset every time Render redeployed - because the free tier uses an ephemeral filesystem.

To persist data, I needed either:

A Render Disk ($0.25/GB/mo, minimum $2.50) — but this only works on paid instances
A Render PostgreSQL database ($7/mo after 90-day free trial)

I upgraded to a paid instance ($7/mo) and added a Render Disk. Now SQLite persists. But my $0/mo "free tier" side project just became $7/mo, and that's before adding a real database.

Time to working URL: ~12 minutes (free tier took 5 min, then 7 more to figure out persistence and upgrade).

Other friction:

Free tier instances sleep after 15 minutes of inactivity. First request after sleeping takes 30-50 seconds to cold-start.
Logs are limited. When my app crashed during one test, the error was truncated.
No SSH on the free tier.

Monthly cost: $7/mo minimum for persistent storage. $14/mo if you need a proper database.

Verdict: Good platform, but the free tier is misleading for anything with a database. You'll upgrade quickly.

4. DigitalOcean Droplet — Full Control, Full Responsibility

The classic approach. Rent a $6/mo VPS and set everything up yourself.

What happened:

Created a $6/mo droplet (1 vCPU, 1 GB RAM, 25 GB SSD). SSH'd in. Then spent the next 40 minutes:

apt update && apt install nodejs npm nginx certbot — 3 min
Uploaded my code via scp — 1 min
npm install — 1 min
Created a systemd service file so the app starts on boot — 5 min (had to Google the syntax)
Configured nginx as a reverse proxy — 8 min (always forget the proxy_pass format)
Set up a domain and DNS A record — 5 min (waiting for propagation)
Ran certbot for SSL — 3 min
Configured UFW firewall — 3 min
Debugged why nginx was returning 502 — 10 min (forgot to open port 3000 internally)
Tested everything — 2 min

Time to working URL: ~47 minutes.

My app runs great now. Full SSH access, real filesystem, SQLite works perfectly. I can install anything, run a database, store files, no limitations.

The trade-off: I'm now responsible for everything. OS updates, security patches, SSL certificate renewals, monitoring, backups. If the app goes down at 2 AM, that's my problem.

Monthly cost: $6/mo (just the droplet).

Verdict: Best value if you know Linux. Worst experience if you just want to ship something fast. 47 minutes of DevOps for an app I built in 10 minutes with AI.

5. InstaPods — One Command

Full disclosure: I built InstaPods. I'm including it because the comparison is useful and I'll be honest about the gaps.

What happened:

$ cd expense-tracker
$ instapods deploy

  Deploying expense-tracker
  Detected nodejs (package.json)

  Creating pod ·················· done  1.2s
  42 files uploaded ············· done  0.8s
  Reloading ····················· done  1.4s
    4 deps installed · service active · HTTP 200

  Deployed in 3.4s
  https://expense-tracker.instapods.app

The CLI detected Node.js from package.json, created a pod, uploaded code, installed dependencies, started the app. SQLite works because it's a real Linux server with a persistent filesystem.

Time to working URL: ~2 minutes (including CLI install).

Monthly cost: $3/mo (Launch plan).

The honest gaps:

New platform. Launched in 2026, smaller community than Railway or Render
Single region right now (EU - Nuremberg). Add ~120ms latency from US East, ~250ms from Asia
No built-in metrics dashboard or log streaming (SSH in and check manually)

Risk reducers: $10 free credit on signup (no credit card required). Flat $3/mo - no usage spikes. SSH access on every plan, so you're never locked in.

Verdict: Fastest deploy by far. Cheapest managed option. But you're betting on a newer platform.

What I Actually Learned

1. Serverless platforms can't handle vibe-coded apps. AI tools generate full-stack apps with persistent storage, not static sites. Vercel and Netlify are the default recommendations on Reddit, but they break the moment your app needs to write to disk. SQLite, file uploads, background jobs - none of it works on serverless.

2. "Free tier" disappears the moment you need a database. Render's free tier, Vercel's hobby plan, Railway's trial - they work for static demos. The moment your app needs persistent data (which is almost always with AI-generated apps), you're paying $7-20/mo.

3. Raw VPS setup takes 4x longer than building the app. DigitalOcean works perfectly - if you're willing to spend 47 minutes on nginx, SSL, systemd, and firewall setup for an app you built in 10 minutes. For someone who just vibe-coded their first app, it's a wall.

4. The deploy gap is now the hardest part of building with AI. My expense tracker took 10 minutes to build and anywhere from "impossible" (Vercel) to 47 minutes (DigitalOcean) to deploy. AI closed the build gap. The deploy gap is still wide open.

The Comparison, One More Time

	Vercel	Railway	Render	DigitalOcean	InstaPods
Works for full-stack?	No	Yes	Yes (paid)	Yes	Yes
SQLite works?	No	Yes	Paid only	Yes	Yes
SSH access?	No	No	Paid only	Yes	Yes
Setup time	Failed	~8 min	~12 min	~47 min	~2 min
Monthly cost	-	$5-15	$7-14	~$6	$3
You manage	-	Nothing	Nothing	Everything	Nothing
Billing model	-	Usage	Tier + usage	Flat	Flat

Who Should Use What

If your app is frontend-only (static HTML/CSS/JS, no backend):
Vercel or Netlify. They're free and they're great at what they do.

If you want a mature PaaS and don't mind usage-based billing:
Railway. Best developer experience of the PaaS options. Solid team, good docs, active community.

If you want full control and know Linux:
A Hetzner or DigitalOcean VPS at $4-6/mo. You'll spend time on setup, but you own every layer.

If you want the fastest path from "works on localhost" to "works on the internet":
InstaPods. One command, $3/mo, real server. I built it specifically for this use case.

FAQ

Can I deploy a full-stack AI-generated app to Vercel?
Only if you rewrite the backend for serverless. Vercel runs Node.js as stateless functions - no persistent filesystem, no SQLite, no long-running processes. If your AI tool generated a standard Express/Flask/Django app, Vercel won't work without restructuring.

What's the cheapest way to deploy a vibe-coded app?
For managed hosting: InstaPods at $3/mo or Railway at $5/mo minimum. For DIY: a Hetzner VPS at ~$3.50/mo, but you'll spend 30-60 minutes on setup. For static-only apps: Vercel and Netlify are free.

Do AI-generated apps need special hosting?
No, but they need hosting that supports whatever the AI generated. Most AI coding tools produce full-stack apps with backends and file-based databases (SQLite). That rules out serverless platforms. You need a host with a persistent filesystem and the ability to run a long-lived process.

Can Claude Code or Cursor deploy directly?
Yes, through MCP (Model Context Protocol). InstaPods has an MCP server that lets AI coding tools create pods, push code, and manage apps directly from the conversation. No terminal needed.

For more comparisons, I wrote about the cheapest way to self-host n8n, Coolify vs Cloudron vs CapRover, and a 6-platform comparison for self-hosted apps.

AI tools closed the build gap. The deploy gap is still wide open. Until it catches up, "how do I get this online?" will keep being the hardest part of vibe coding.

What's your deploy workflow for AI-generated apps? I'm curious what other people are running. Drop it in the comments.

The Cheapest Way to Self-Host n8n in 2026

Vikas Singhal — Sun, 08 Mar 2026 14:41:24 +0000

Last updated: March 2026

n8n is one of the best open source automation tools out there - a self-hosted alternative to Zapier and Make that gives you full control over your workflows and data. But "self-hosted" means you need somewhere to run it, and finding the cheapest way to self-host n8n without sacrificing reliability takes some research.

I've been running n8n for the past year and I've tried more hosting setups than I'd like to admit. Here's every option I found, ranked by actual monthly cost, with the trade-offs that matter.

TLDR: The cheapest managed option is InstaPods at $3/mo or PikaPods at ~$3.80/mo - both handle everything. The cheapest self-managed option is a $4-5/mo Hetzner VPS with Docker Compose. n8n Cloud (the official hosted version) starts at $24/mo. The right answer depends on whether your time or your money is more expensive.

Every Way to Host n8n, Ranked by Cost

Method	Monthly Cost	You Manage	Setup Time
Oracle Cloud free tier	$0	Everything	~1 hr
InstaPods	$3	Nothing	~30 sec
PikaPods	~$3.80	Nothing	~1 min
Hetzner VPS + Docker Compose	~$4-5	Everything	~45 min
Coolify on Hetzner VPS	~$5-8	VPS + OS	~15 min
CapRover on Hetzner VPS	~$5-6	Everything	~30 min
Railway	~$5-15	Nothing	~5 min
Elestio	~$17	Nothing	~3 min
Cloudron on Hetzner VPS	~$21	VPS + OS	~40 min
n8n Cloud (official)	$24+	Nothing	~2 min

Let me walk through each one.

1. Hetzner VPS + Docker Compose (~$4-5/mo)

The classic DIY approach. Rent a cheap VPS, install Docker, run n8n with docker-compose.

version: "3"
services:
  n8n:
    image: n8nio/n8n
    ports:
      - "5678:5678"
    environment:
      - N8N_BASIC_AUTH_ACTIVE=true
      - N8N_BASIC_AUTH_USER=admin
      - N8N_BASIC_AUTH_PASSWORD=changeme
    volumes:
      - n8n_data:/home/node/.n8n
volumes:
  n8n_data:

Cost breakdown:

Hetzner CX22: EUR 3.99/mo (~$4.30) - 2 vCPU, 4 GB RAM, 40 GB storage
Or Hetzner CAX11 (ARM): EUR 3.29/mo (~$3.55) - 2 Ampere vCPU, 4 GB RAM

What you also need to set up:

Reverse proxy (nginx or Caddy) for HTTPS
SSL certificate (Let's Encrypt + certbot)
Firewall rules (UFW)
Automatic updates for n8n and the OS
Backups (cron job + rclone to S3/Backblaze)
Monitoring (so you know when it goes down)

Pros: Cheapest paid option. Full control. Can run other apps on the same server.
Cons: You're the sysadmin. Every component is your responsibility. Budget 30-60 minutes for initial setup, plus ongoing maintenance.

Best for: Developers comfortable with Linux, Docker, and nginx who want maximum control.

2. Oracle Cloud Free Tier ($0/mo)

Oracle offers an always-free ARM instance (4 OCPU, 24 GB RAM) that can run n8n. It's genuinely free - not a trial.

Cost: $0 if you can get an instance. Availability is limited and instances are hard to provision (you'll get "out of capacity" errors for days or weeks).

Catch: Oracle's free tier has been reliable for some people and randomly terminated for others. There are reports of free-tier instances being reclaimed without warning. I wouldn't run production workflows on it.

Best for: Testing n8n or running non-critical personal automations. Not recommended for anything you depend on.

3. Coolify on Hetzner VPS (~$5-8/mo)

Install Coolify (free, open source) on a VPS and deploy n8n through its one-click catalog.

Cost breakdown:

Hetzner CX22: ~$4.30/mo (if running Coolify + n8n on same server)
Hetzner CX32: ~$7.50/mo (recommended - Coolify itself needs 2 GB RAM)

Pros: Web UI for managing n8n. Auto-updates. Easier than raw Docker Compose. Can run other apps alongside n8n.
Cons: Coolify needs 2 vCPU + 2 GB RAM minimum. On a small server, that's most of your resources gone before n8n even starts. Builds compete with running apps for CPU.

Best for: People who want a web dashboard but don't want to pay for Cloudron's license.

4. CapRover on Hetzner VPS (~$5-6/mo)

Install CapRover (free) and deploy n8n from its template catalog.

Cost: Same as DIY Docker Compose - just the VPS cost. CapRover is free.

Pros: One-click n8n install. Web UI. Git-based deploys for custom apps.
Cons: CapRover development has stalled (last commit Dec 2025). Docker Swarm dependency is a long-term risk. You still manage the VPS, OS, backups.

Best for: Existing CapRover users. I'd pick Coolify for new setups.

5. InstaPods ($3/mo)

InstaPods has n8n as a one-click app. Click deploy, get a running instance with SSL in about 30 seconds. $3/mo on the Launch plan.

Full disclosure: I built InstaPods. Including it because it's genuinely the cheapest managed option and the comparison is useful. I'll be honest about the gaps.

Cost: $3/mo (Launch plan: 0.5 vCPU, 512 MB RAM, 5 GB storage). $7/mo for the Build plan (1 vCPU, 1 GB RAM) if your workflows need more power.

Pros: Fastest setup of any option here. No server to manage. SSL and URL included. $10 free credit on signup.
Cons: New platform (launched 2026). Single region (EU - Nuremberg). Smaller community than established platforms. The $3 plan works for light workflows but heavier automations need the $7 plan.

Best for: People who want n8n running in under a minute with zero maintenance.

6. PikaPods (~$3.80/mo)

PikaPods is managed hosting for open source apps. Pick n8n, set your resource sliders, done.

Cost: ~$3.80/mo for n8n with default resources. Adjustable up based on CPU/RAM/storage needs.

Pros: Zero maintenance. Revenue sharing with n8n (supports the project). $5 welcome credit. Established platform with a good track record.
Cons: Limited configuration options compared to self-hosted. Can't install custom nodes that need system packages.

Best for: Non-technical users who want managed n8n at the lowest price.

7. Elestio (~$17/mo)

Elestio is fully managed hosting. Pick n8n, pick a cloud provider, they deploy it on a dedicated VM.

Cost: ~$17/mo minimum (Hetzner-backed VM). Higher for other cloud providers.

Pros: True zero-maintenance. Auto-updates, backups, security patches all handled. Choose from 8 cloud providers.
Cons: One app = one VM. $17/mo for n8n alone is expensive when you can get managed hosting for $3-4/mo elsewhere.

Best for: Teams with budget who want fully managed hosting on their preferred cloud provider.

8. Cloudron on Hetzner VPS (~$21/mo)

Install Cloudron on a VPS and deploy n8n from the app store. Best admin UX of any self-hosted platform.

Cost: VPS (~$5/mo) + Cloudron license (EUR 15/mo) = ~$21/mo.

Pros: Best admin dashboard. SSO, email, backups built in. Rock-solid updates.
Cons: EUR 15/mo license makes it the most expensive option for running a single app. Makes more sense if you're running 5-10 apps on one server.

Best for: People already running Cloudron for other apps. Not worth the license fee for n8n alone.

9. Railway (~$5-15/mo)

Railway is a PaaS that can run n8n via Docker image.

Cost: Usage-based. ~$5/mo for light usage, can spike to $15+ with active workflows. Unpredictable.

Pros: Quick setup. No server management. Nice developer UX.
Cons: Usage-based billing means costs are hard to predict. No SSH access. Sleeps on hobby plan.

Best for: Developers already using Railway who want to add n8n to their stack.

10. n8n Cloud ($24+/mo)

The official hosted version from the n8n team.

Cost: Starter at $24/mo (2,500 workflow executions). Pro at $60/mo (10,000 executions). Enterprise pricing above that.

Pros: Officially supported. Always up to date. Built-in collaboration features. No infrastructure to manage.
Cons: Most expensive option by far. Execution limits - you pay more as workflows run more. At scale, self-hosting saves thousands per year. I wrote a detailed n8n pricing breakdown if you want the full per-execution math.

Best for: Teams that need official support, collaboration features, and don't mind paying premium pricing.

The Real Comparison

Here's what it comes down to:

If your time is worth more than $5-15/mo:
Go managed. PikaPods (~$3.80/mo) and InstaPods ($3/mo) are the cheapest managed options. Both handle updates, backups, and SSL. You get n8n running in under a minute.

If you want full control and have Linux skills:
Hetzner VPS + Docker Compose at ~$4-5/mo. You manage everything, but you have SSH access, can run other apps on the same server, and can customize anything.

If you want a middle ground:
Coolify on a Hetzner VPS (~$5-8/mo). You get a web dashboard for managing n8n and other apps, with auto-updates. Still need to maintain the VPS itself.

If you need official support:
n8n Cloud ($24-60/mo). Most expensive, but it's the only option with official n8n team support and collaboration features.

FAQ

How much RAM does n8n need?
For light usage (under 20 workflows, basic automations): 512 MB is enough. For moderate usage (50+ workflows, HTTP triggers, database operations): 1 GB minimum. For heavy usage (hundreds of workflows, large data processing): 2-4 GB.

Can I run n8n on a Raspberry Pi?
Yes. n8n has ARM Docker images. A Raspberry Pi 4 with 4 GB RAM runs n8n fine for personal use. Performance drops with complex workflows or high-frequency triggers.

Is self-hosted n8n missing features compared to n8n Cloud?
Self-hosted n8n has all the core workflow features. n8n Cloud adds collaboration (multiple users editing workflows), execution history with longer retention, and official support. For solo use, self-hosted is feature-complete.

How do I back up n8n?
n8n stores data in ~/.n8n (SQLite by default). Back up that directory. On Docker, back up the volume. On managed platforms (PikaPods, InstaPods, Elestio), backups are handled for you.

Can I migrate from n8n Cloud to self-hosted?
Yes. Export your workflows as JSON from n8n Cloud, then import them into your self-hosted instance. Credentials need to be re-entered manually (they don't export for security reasons).

Should I use SQLite or PostgreSQL with n8n?
SQLite is fine for most self-hosted setups. Switch to PostgreSQL if you have 100+ active workflows or need high-frequency triggers (multiple per second). For a $3-5/mo hosting setup, SQLite keeps things simple.

For a broader comparison of self-hosted platforms (not just for n8n), I wrote about Coolify vs Cloudron vs CapRover and a 6-platform comparison.

What's your n8n hosting setup? I'm curious what other people are running. Drop it in the comments.

Coolify vs Cloudron vs CapRover in 2026: I Self-Hosted Apps on All Three

Vikas Singhal — Sun, 08 Mar 2026 14:31:16 +0000

Last updated: March 2026

Coolify, Cloudron, and CapRover are the three most popular platforms for self-hosting open source apps on your own server. They all promise one-click app installs, automatic SSL, and a web dashboard to manage everything.

I've used all three to run n8n, Uptime Kuma, and Plausible over the past year. Each one solved problems the others didn't, and each one had trade-offs I only discovered after committing. Here's what actually matters when choosing between them - with real setup times, real costs, and the problems you'll hit that the landing pages don't mention.

TLDR: Coolify is the best all-around choice in 2026 - free, actively maintained, 280+ apps, connects to unlimited servers. Cloudron has the most polished UX with built-in SSO and email, but costs EUR 15/mo on top of your VPS. CapRover is free and has the biggest app catalog (346), but development has stalled since December 2025. If you don't want to manage a VPS at all, managed platforms like PikaPods (~$2-4/mo) or InstaPods ($3/mo) handle everything.

Quick Comparison

	Coolify	Cloudron	CapRover
Price	Free (self-hosted) or $5/mo (cloud)	EUR 15/mo license + VPS	Free + VPS
App catalog	280+	100-140	346
Setup time	~5 min	~30 min	~20 min
Min server	2 vCPU, 2 GB RAM	1 vCPU, 1 GB RAM	1 vCPU, 1 GB RAM
Architecture	Docker via SSH	Docker on host	Docker Swarm
Custom apps	Yes (Git deploy)	Yes (custom packages)	Yes (Git/Docker)
Auto updates	Yes	Yes	No
Built-in email	No	Yes	No
SSO	No	Yes	No
Active development	Very active	Active	Slow (last commit Dec 2025)
Multi-server	Yes (unlimited)	1 per license	Yes (Docker Swarm)

Now let me break each one down.

Coolify - The Modern Choice

Coolify is the newest of the three and the most actively developed. It's an open source platform that connects to your servers via SSH and deploys apps using Docker. You can self-host it for free or use their cloud version for $5/mo.

Setup Experience

I installed Coolify on a fresh Hetzner VPS (CX22, $5.39/mo) in about 5 minutes:

curl -fsSL https://cdn.coollabs.io/coolify/install.sh | bash

The installer handles Docker, sets up the database, and gives you a web UI. First impression: the dashboard looks modern and feels like a real product.

What's Good

Free and open source. The self-hosted version includes every feature. No license fees.
280+ one-click services. n8n, Uptime Kuma, Plausible, Ghost, Gitea - most popular apps are there.
Connects to unlimited servers. Add remote servers via SSH. One Coolify instance manages all of them.
Git-based deploys. Connect GitHub/GitLab repos for custom apps. Builds with Nixpacks or Dockerfile.
Auto-updates for apps. Set it and forget it. Apps update on your schedule.
Very active development. Andras (the creator) ships updates almost daily. The GitHub is one of the most active in the self-hosted space.

What's Not Great

Resource hungry. Coolify itself needs 2 vCPU and 2 GB RAM minimum. On a small server, that's half your resources gone before you install a single app.
Builds compete with apps. If you build and run apps on the same server, builds eat CPU/memory that your running apps need. I had n8n slow to a crawl during a Plausible rebuild.
Rough edges. Occasional UI bugs, unclear error messages when deploys fail. It's improving fast, but it's not as polished as Cloudron.
No built-in email or SSO. If you need unified login across apps or email sending, you'll set that up separately.

Real Cost

Coolify server: ~$5/mo (Hetzner CX22)
If you run apps on the same server: $5/mo total
If you use a separate server for apps (recommended): $10-11/mo total

Cloudron - The Polished One

Cloudron is the most mature of the three. It installs on your VPS and gives you what I'd describe as an "iOS App Store for your server" - a clean UI where you browse apps, click install, and everything just works.

Setup Experience

Setup takes longer than Coolify - about 30 minutes. You need:

A VPS running Ubuntu 24.04 (KVM only - no LXC, no OpenVZ)
A domain with wildcard DNS pointing to your server
Run the install script

The DNS requirement tripped me up and added 15 minutes of troubleshooting. You can't use an IP address - Cloudron needs *.yourdomain.com pointing to the server before installation completes.

What's Good

Best admin UX. No contest. The dashboard is clean, responsive, and everything works like you'd expect. Updates, backups, user management - all polished.
Built-in email server. Cloudron runs a full mail server. Your apps can send email without configuring SMTP.
SSO across apps. One login for all apps. Add team members, set permissions per app.
Rock-solid updates. App updates are tested and staged. I've never had an update break a running app.
Good documentation. Clear, complete, with troubleshooting for common issues.

What's Not Great

License costs EUR 15/mo. On top of your VPS cost. The free tier limits you to 2 apps - barely enough to test a real setup.
Smaller catalog. 100-140 apps vs 280+ on Coolify and 346 on CapRover.
KVM only. No ARM servers, no LXC containers, no OpenVZ. This rules out some cheaper VPS providers.
Wildcard DNS required. You need a domain. No IP-only option.
1 server per license. Want to run apps on a second server? That's another EUR 15/mo.

Real Cost

VPS: ~$5-6/mo (Hetzner CX22)
Cloudron license: EUR 15/mo (~$16)
Total: ~$21-22/mo for unlimited apps on one server

The per-app cost drops fast if you run 5-10 apps on one server. But for 1-2 apps, it's the most expensive option here.

CapRover - The OG

CapRover has been around the longest. It's free, open source, and runs on Docker Swarm. Think of it as "build your own Heroku" - deploy via Git push, Dockerfile, or one-click templates.

Setup Experience

Setup takes about 20 minutes. Install Docker, run CapRover's setup, then configure DNS and SSL through the web UI:

docker run -p 80:80 -p 443:443 -p 3000:3000 \
  -e ACCEPTED_TERMS=true -v /var/run/docker.sock:/var/run/docker.sock \
  -v /captain:/captain caprover/caprover

The initial setup wizard walks you through DNS and SSL configuration. Straightforward if you're comfortable with Docker.

What's Good

Completely free. No license, no cloud fee, just your VPS cost.
346 one-click app templates. The biggest catalog of the three.
Deploy custom apps via Git. Push to CapRover's Git endpoint and it builds with your Dockerfile or Captain Definition file.
Multi-node clusters. Docker Swarm support for running apps across multiple servers.
Mature and stable. It's been around for years. Most edge cases have been found and documented.

What's Not Great

Development has stalled. Last commit was December 2025. No updates in 3+ months. Open issues are piling up. This is the biggest red flag.
You're the sysadmin. OS updates, Docker upgrades, security patches, backups - all on you.
Docker Swarm risk. Docker deprecated Swarm mode. CapRover's architecture depends on it.
Outdated templates. Many one-click apps use old versions. Some templates are broken or unmaintained.
Docker version sensitivity. A Docker Engine update in late 2025 broke CapRover on existing installs. If you're running it, you need to be careful with system updates.
No auto-updates for apps. You manually update each app through the UI.

Real Cost

VPS: ~$5-6/mo (Hetzner CX22)
Total: ~$5-6/mo for unlimited apps

Cheapest option by far - if your time is free.

Feature-by-Feature Comparison

App Deployment

All three let you install apps with one click and deploy custom projects via Git or Docker. Coolify and CapRover auto-detect your framework with Nixpacks or a Dockerfile. Cloudron uses its own packaging format, so app support depends on the Cloudron team packaging it first.

Winner: Coolify - largest selection of modern apps with auto-detection for custom projects.

Server Management

Cloudron manages everything on one server with a clean UI. Coolify connects to remote servers via SSH - add as many as you want. CapRover supports Docker Swarm clusters but requires manual node setup.

Winner: Coolify for multi-server. Cloudron for single-server simplicity.

Updates & Maintenance

Cloudron handles app updates, OS patches, and backups automatically. Coolify handles app updates but you manage the OS. CapRover doesn't auto-update anything.

Winner: Cloudron - the most hands-off experience.

User Management

Cloudron has built-in SSO, email, and per-app permissions. Coolify has basic team support. CapRover has no user management beyond the admin account.

Winner: Cloudron - not even close.

Reliability

Cloudron's app packaging is the most stable. Coolify moves fast and occasionally ships bugs. CapRover's stalled development means bugs stick around.

Winner: Cloudron for stability. Coolify for active development momentum.

What About Managed Alternatives?

All three platforms above require you to manage your own VPS. You provision the server, install the platform, and handle OS-level maintenance. If you'd rather skip the server management entirely, there are managed alternatives worth knowing about:

PikaPods - Managed hosting for 200+ open source apps. Pick an app, set resources, done. Starting at ~$2-4/mo per app. No server to manage at all.
Elestio - Fully managed with 400+ apps. Each app gets its own VM (~$17/mo per app). Zero maintenance.
InstaPods - One-click app hosting for a growing catalog (n8n, Uptime Kuma, Beszel, and more). $3/mo flat per app, no server management. Full disclosure: I built this one. (For a detailed comparison of all 6 platforms, I wrote a separate deep-dive.)

The managed platforms cost slightly more but save real time. If you're running 1-2 apps and don't enjoy sysadmin work, managed hosting often makes more sense than running Coolify or CapRover yourself.

Decision Guide

Choose Coolify if:

You want free, modern, and actively maintained
You're comfortable managing a VPS
You run apps across multiple servers
You deploy custom apps from GitHub repos

Choose Cloudron if:

You want the most polished, hands-off experience on a single server
You need SSO, email, or team user management
You're running 5+ apps (cost per app drops fast)
Stability matters more than cutting-edge features

Choose CapRover if:

You want the cheapest option and don't mind maintaining it yourself
You need Docker Swarm multi-node clusters
Your apps are in their template catalog
You're comfortable with the risk of a stalling project

Choose a managed platform (PikaPods, Elestio, InstaPods) if:

You don't want to manage a VPS at all
You're running 1-3 apps and want zero ops work
Your time is worth more than the $5-15/mo difference

FAQ

Which is better for beginners - Coolify or Cloudron?
Cloudron. The UI is more intuitive, updates are automatic, and you're less likely to break something. Coolify is more powerful but assumes more Linux/Docker knowledge. If you've never managed a VPS before, Cloudron's license fee is worth the smoother experience.

Can I migrate between these platforms?
Not easily. Each platform stores app data differently. Moving from CapRover to Coolify means backing up your app's data (database dumps, file exports), setting up the app fresh on the new platform, and restoring the data. There's no one-click migration tool between any of them.

Do I need a powerful server?
For 1-3 lightweight apps (Uptime Kuma, Bookstack, Gitea): 1 vCPU, 2 GB RAM is enough on Cloudron or CapRover. Coolify itself needs 2 GB, so budget 4 GB total. For heavier apps like n8n with many workflows or Plausible with real traffic: 2 vCPU, 4 GB RAM minimum on any platform.

Is CapRover dead?
Not dead, but development has slowed significantly. The last commit was December 2025, and Docker Swarm (which CapRover depends on) is in maintenance mode. It still works fine for existing installs, but I'd hesitate to start a new setup on it in 2026 when Coolify offers a similar feature set with active development.

What's the cheapest way to self-host n8n?
Self-managed: Coolify on a $5/mo Hetzner VPS. Total cost: $5/mo, but you manage the server. Managed: PikaPods at ~$3.80/mo or InstaPods at $3/mo - both handle everything for you.

Can I run these platforms on a Raspberry Pi?
CapRover supports ARM and can run on a Pi 4. Coolify has experimental ARM support. Cloudron does not support ARM at all - KVM/x86 only.

I wrote a more detailed comparison including Elestio, PikaPods, and InstaPods here, and guides for deploying without nginx and deploying without DevOps knowledge if you're looking for simpler alternatives to managing your own server.

What are you self-hosting, and which platform are you using? Drop your setup in the comments.