remote-system-maintenance

Remote system maintenance

Structured procedures for diagnosing and maintaining remote Linux systems via SSH/tmux.

Installation

/plugin install remote-system-maintenance@2389-research

What this plugin does

Walks you through system diagnostics and cleanup on remote Linux boxes, focused on Ubuntu/Debian. Three diagnostic phases, a seven-stage cleanup sequence, and documentation templates so you actually record what you did. Expect to recover 2+ GB in a thorough session.

When to use

• Performing system maintenance on remote Linux servers
• Recovering disk space on Ubuntu/Debian systems
• Running diagnostics on remote systems
• Cleaning up package caches, journals, or snap revisions

Quick example

# Phase 1: Initial diagnostics
hostname && df -h && free -h && uptime
ps aux | head -20
ps aux | awk '$8 ~ /Z/ {print}'  # Zombie detection

# Phase 2: Log review
journalctl -p err -n 50
journalctl --disk-usage

# Phase 3: Cleanup sequence
apt update && apt upgrade -y
apt autoremove -y && apt clean
journalctl --vacuum-time=7d

# Snap cleanup (biggest wins!)
snap list --all | awk '/disabled/{print $1, $3}'
snap remove package-name --revision=123

# Document: hostname, before/after disk, MB freed per category

Three-phase approach

Phase 1: Initial diagnostics

Capture baseline system state:

• Hostname and system identification
• Resource utilization (disk, memory, CPU)
• Process status and load
• Zombie process detection

Phase 2: System log review

Examine system health:

• Recent error messages in system logs
• Journal disk consumption
• Critical service status
• Authentication and security events

Phase 3: Package assessment

Find maintenance opportunities:

• Upgradable packages
• Orphaned configurations
• Unused dependencies
• Package cache size

Ubuntu/Debian cleanup sequence

Run these seven stages in order:

1. Package cache refresh -- apt update
2. System upgrades -- apt upgrade
3. Orphan removal -- apt autoremove
4. Cache purge -- apt clean
5. Journal pruning -- journalctl --vacuum-time=7d
6. Snap revision cleanup -- remove disabled snap revisions
7. Temporary directory check -- review /tmp and /var/tmp

Snap revision cleanup

Snap keeps old revisions by default. This is where the big wins are:

# List all disabled snap revisions
snap list --all | awk '/disabled/{print $1, $3}'

# Remove specific revision
snap remove <package-name> --revision=<revision-number>

You have to remove each revision by number explicitly.

Expected results

Typical recovery per category:

• Journal vacuuming: 300-600 MB
• Snap revision cleanup: 500 MB to 2 GB
• Package cache purge: 100-500 MB
• Total: 2+ GB in a comprehensive session

Documentation requirements

Every maintenance session should produce a structured log covering:

1. System identification -- hostname, OS version, kernel, operator
2. Resource states -- initial and final disk/memory/CPU usage
3. Actions taken -- commands executed, MB/GB freed per category
4. Follow-up recommendations -- remaining issues, future needs

Time commitment

A typical maintenance session takes 15-30 minutes including diagnostics, cleanup, and documentation.

Documentation

See skills/SKILL.md for the complete maintenance procedures.

Philosophy

Structure ad-hoc operational work with checklists and documentation. Quantify everything.