Optimization Tips, Auto-Cleanup & Real-World Troubleshooting for n8n Self-hosted

Author: Vinh Automation

Level: Intermediate

Applies to: n8n systems on Ubuntu VPS already installed via Part 1

---

Part 1: Overview - The “Nightmares” of Running n8n in Production

You have successfully installed n8n Self-hosted following the previous guide. Congratulations!

However, getting it installed is only half the journey. Running it in production on a modest VPS always comes with hidden “nightmares”:

Issue	Symptoms	Consequences
Docker log files bloating	Disk full after 2-3 months	n8n stops working, cannot write data
RAM overflow from heavy workflows	VPS freezes, cannot SSH in	Workflow data lost before it can be saved
Cloudflare Tunnel UDP throttling	502 Bad Gateway / Error 1033	Users cannot access n8n
Dangling Docker images	Disk silently fills up	Reduced performance, wasted space

This article is a real-world experience summary — every error has happened on a real system and has been thoroughly resolved. You will get a complete set of weapons to keep n8n running rock-solid all year round.

---

Part 2: 5 Optimization & Cleanup Weapons

🗡️ Weapon #1: Optimize MobaXterm - Access VPS in One Click

Every time you open MobaXterm and have to type your SSH password, it wastes time. Instead:

1. In the left Sessions column, right-click your VPS session → Edit session.
2. Look at the Username field (root), click the person icon next to the golden key.

Click the person icon next to the Username field to open Password Settings

3. Click New (blue pen icon) → enter root / your VPS password → OK.

🗡️ Weapon #2: Enable SWAP - Prevent Silent Crashes When n8n is Overloaded

When AI workflows process large amounts of data, n8n can eat up all your RAM. Linux has an OOM Killer that will… kill n8n to protect the system! SWAP (virtual RAM) is your lifesaver:

# Create a 2GB swap file, set permissions, activate, and write to fstab
sudo fallocate -l 2G /swapfile && \
  sudo chmod 600 /swapfile && \
  sudo mkswap /swapfile && \
  sudo swapon /swapfile && \
  echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab

Detailed explanation:

Command	Purpose
fallocate -l 2G	Create an empty 2GB file - faster than using dd.
chmod 600	Only root can read/write - security measure.
mkswap	Format the file as a swap partition.
swapon	Activate swap immediately.
tee -a /etc/fstab	Write to fstab so swap auto-activates on every reboot.

Verify: run free -h - if you see Swap: 2.0Gi, it worked.

Also, set the timezone so Cron/Schedule nodes in n8n run at the correct time:

sudo timedatectl set-timezone Asia/Ho_Chi_Minh

Verify: run date - should show +07.

🗡️ Weapon #3: Limit Docker Logs - Prevent Hidden Disk Full Errors

By default, Docker keeps logs indefinitely. After 3 months, log files can bloat to 10-20GB and cripple your disk.

Solution: Remove the old container and re-run with log limits:

# Remove old container (workflow data is safe in the volume)
docker rm -f n8n_app

# Re-run with 10MB per log file, max 3 files
docker run -d \
  --name n8n_app \
  --restart always \
  -p 5678:5678 \
  --log-driver json-file \
  --log-opt max-size=10m \
  --log-opt max-file=3 \
  -v n8n_data:/home/node/.n8n \
  n8nio/n8n:latest

Important log parameters:

Parameter	Value	Meaning
--log-driver json-file	json-file	Docker’s default log format.
--log-opt max-size=10m	10m	Each log file maxes out at 10MB before rolling.
--log-opt max-file=3	3	Keep only the 3 most recent log files.

Verify: docker inspect n8n_app | grep -A 5 LogConfig

🗡️ Weapon #4: Fix 502 Bad Gateway & Error 1033 (Caused by ISP UDP Throttling)

Symptoms

You visit https://n8n.your-domain.com and see:

• 502 Bad Gateway (Host Error) from Cloudflare — or
• Error 1033: Argo Tunnel error

But the VPS still responds to ping, and SSH works fine.

Diagnosis

Check the Tunnel logs:

docker logs cloudflare_tunnel --tail 20

The critical error line:

ERR Failed to dial a quic connection error="failed to dial to edge with quic: timeout: no recent network activity"
INF precheck component="TCP Connectivity" details="HTTP/2 connection successful"

Root Cause

By default, Cloudflare Tunnel uses the QUIC protocol (running on UDP ports) to connect. However, many ISPs throttle or block UDP ports going internationally. Result: the Tunnel times out constantly, while the TCP path works perfectly.

4-Step Fix Process

Step 1: Delete the old Tunnel - clean up Connectors

Go to Cloudflare Zero Trust → Networks → Tunnels:

• Click the failing Tunnel → the 3 dots → Delete tunnel.

Step 2: Create a new Tunnel & Copy the Token

• Click Create a tunnel → give it a new name → select Docker.
• Copy the long token string eyJh... after --token.

Step 3: Delete the old DNS record

Open a new tab → go to Cloudflare Dashboard → DNS:

• Find the CNAME record named n8n (from the old Tunnel) → Delete.

Step 4: Run the new Tunnel - force TCP (HTTP/2)

# Remove the old tunnel container
docker rm -f cloudflare_tunnel

# Run the new tunnel with --network=host and force http2 protocol
docker run -d \
  --name cloudflare_tunnel \
  --network=host \
  --restart always \
  cloudflare/cloudflared:latest \
  tunnel --protocol http2 run --token <YOUR_NEW_TOKEN>

What changed:

Parameter	First setup (Part 1)	This fix	Why?
--network=host	❌ Not used	✅ Used	Tunnel shares the VPS network stack, improving connection speed.
--protocol http2	❌ Not used (defaults to QUIC)	✅ Used	Forces Tunnel to use TCP instead of UDP — avoids ISP throttling.

Verify with the logs:

docker logs cloudflare_tunnel --tail 10

Expected output:

SUMMARY: Environment ready with degraded transport. cloudflared will proceed using 'http2'.

After that, the browser should work normally.

🗡️ Weapon #5: Auto-Cleanup Script - Sunday Cronjob

Ubuntu and Docker systems running for a long time accumulate:

• apt cache files.
• Unnecessary update packages.
• Docker dangling images — untagged images taking up GB.

Solution: Set up a crontab to auto-clean every Sunday at 00:00.

Open crontab:

crontab -e

Type 1 to select 1. /bin/nano

Scroll to the bottom of the file, paste this exact line:

0 0 * * 0 apt-get autoremove -y && apt-get clean && journalctl --vacuum-time=7d && docker image prune -f && docker volume prune -f && truncate -s 0 /var/lib/docker/containers/*/*-json.log

crontab -e screen with the weekly cleanup command — press Ctrl+O → Enter to save, Ctrl+X to exit

Press Ctrl + O → Enter to save → Ctrl + X to exit.

Crontab breakdown:

Component	Meaning
0 0 * * 0	Run at 00:00, every Sunday (0 = Sunday).
apt-get autoremove -y	Remove packages that are no longer needed.
apt-get clean	Delete .deb cache files in /var/cache/apt/archives/.
docker system prune -f	Remove all dangling images, unused containers, and networks.
truncate -s 0 /var/lib/docker/containers/*/*-json.log	This “trims” the text log file size (where lines like “Node X executed successfully” are stored) down to 0 to free up disk space, without deleting the log file or container configuration.

Verify the crontab was saved:

crontab -l

You should see the line 0 0 * * 0 apt-get autoremove... displayed.

Standard crontab field breakdown:

• 0: Minute 0
• 0: Hour 0 (Midnight)
• *: Any day of the month
• *: Any month
• 0: Sunday (0 or 7 both mean Sunday)

👉 Every Saturday night at midnight, rolling into Sunday morning, the VPS will quietly trigger this command chain automatically. You don’t need to SSH in manually or reload n8n — everything happens automatically and smoothly. Your system will always be periodically cleaned to run smoothly and reliably.

---

Part 3: Troubleshooting & Best Practices

3.1 Quick Error Reference Table

Error	Cause	Fix command
502 Bad Gateway	Cloudflare Tunnel timeout due to QUIC/UDP	Delete old tunnel, re-run with --protocol http2
Error 1033	Tunnel uses network=host but URL points to container name	Change Public Hostname URL to localhost:5678
n8n won’t start	Port 5678 in use or volume error	docker rm -f n8n_app then re-run the start command
Workflows get slower over time	RAM exhaustion due to missing SWAP	Check free -h, add SWAP if Swap = 0
Cannot SSH into VPS	OOM Killer killed SSH when RAM ran out	Reboot VPS from web hosting, create SWAP immediately

3.2 Quick Health Check Script

Run one command to check the entire system:

echo "=== DOCKER ===" && docker ps -a --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}" && echo "=== DISK ===" && df -h / && echo "=== RAM ===" && free -h && echo "=== SWAP ===" && swapon --show && echo "=== TIME ===" && date && echo "=== CRON ===" && crontab -l

If you see containers Up (not Exited), disk < 80%, swap > 0, and the crontab has the cleanup command — your system is healthy.

3.3 Best Practices

🔒 Security

• Do not store Cloudflare tokens in a text file on the VPS — only use via Docker run command.
• Change your VPS password regularly — especially if using password-based login (not SSH keys).
• Restrict SSH IP: ufw allow from YOUR_IP to any port 22 — only allow your home/office IP to SSH in.

⚡ Performance

• Always enable SWAP — 2GB is the minimum for a 1-2GB RAM VPS.
• Limit Docker logs from day one — don’t wait until the disk is full.
• Delete unused workflows — many inactive workflows still consume RAM when n8n loads.

🏗️ Architecture

• Use Docker volumes (not bind mounts) — easier to backup and migrate.
• Keep a log of past errors — when the same issue happens again, you can look it up faster.
• Upgrade Docker images by specific tag — don’t use :latest in production if you don’t want surprises.

---

Conclusion & Core Lessons

With just a few decisive actions, your n8n system will:

• ✅ Run stably — no silent crashes from RAM or log overflow.
• ✅ Load quickly — no more 502/1033 errors from ISP UDP throttling.
• ✅ Auto-clean — every Sunday, the VPS cleans itself up.
• ✅ Super fast VPS access — double-click via MobaXterm, no password typing.

Core lessons from real-world experience:

1. Don’t waste time looking for a protocol toggle in the Cloudflare Web UI — force it from the Docker command on the VPS (--protocol http2).
2. When using --network=host for the Tunnel to boost speed, always point the target URL to localhost:5678 — never to the container name.
3. Limit Docker logs, enable SWAP, and set up a crontab cleanup — these 3 configurations are mandatory. Set them up right after installing n8n, don’t wait until something breaks.

---

📖 Related articles:

• A-Z Guide: Installing n8n Self-hosted on Ubuntu VPS for Beginners