N8N Server Cleanup (2026-02-25)

Emergency disk cleanup on omelas-hk (68.183.83.159) which was at 95% capacity with only 1.2 GB free.

Before

Actions Taken

1. Removed Unused Docker Images (~5 GB)

Removed images no longer referenced by any container:

• 3 old n8n versions (dangling, superseded by updates)
• nginx-proxy-manager (1.09 GB) — service removed previously
• jlesage/firefox (743 MB) — browser container, unused
• Old python-svc build layers (~1.3 GB)

2. Removed Orphaned Docker Volumes (~100 MB)

Volumes with no associated container:

• firefox-profile (97 MB)
• npm_data + npm_letsencrypt (6 MB)
• n8n_audio_storage (empty)
• nginx-custom-config (empty)

3. Truncated Container Logs (~1.4 GB)

• Postgres container log: 1.2 GB → 0
• Caddy container log: 128 MB → 0
• NocoDB container log: 75 MB → 0

4. Vacuumed Journal Logs (~1.6 GB)

journalctl --vacuum-size=500M

5. Cleared btmp Logs (~837 MB)

> /var/log/btmp
> /var/log/btmp.1

6. Cleared APT Cache (~123 MB)

apt-get clean

7. Configured Docker Log Rotation (prevents recurrence)

// /etc/docker/daemon.json
{
  "log-driver": "json-file",
  "log-opts": {
    "max-size": "10m",
    "max-file": "3"
  }
}

Each container log now caps at 30 MB (3 files x 10 MB).

8. Set Restart Policies

Containers that lacked restart policies were updated:

docker update --restart unless-stopped postgres python-svc n8n-audio-server

After

What Was Preserved

All containers and their associated volumes were kept intact. Only unused images, orphaned volumes, and logs were removed.