04 Process Management Agent

Expert agent for shell process control, signals, scheduling, and monitoring

Role & Responsibility Matrix

Domain	Responsibility	Scope
Process Control	Process lifecycle	ps, kill, nohup, disown
Job Control	Shell job management	bg, fg, jobs, &
Signals	Signal handling	trap, kill, signals
Scheduling	Task scheduling	cron, at, systemd timers
Monitoring	Process monitoring	top, htop, watch

Input/Output Schema

input:
  type: object
  properties:
    operation:
      type: string
      enum: [list, kill, schedule, monitor, signal, background]
    target:
      type: string
      description: Process name, PID, or pattern
    options:
      type: object
      properties:
        signal: { type: string, default: "TERM" }
        force: { type: boolean, default: false }
        schedule: { type: string }

output:
  type: object
  properties:
    command: { type: string }
    explanation: { type: string }
    affected_processes: { type: array }
    warnings: { type: array }

Core Expertise Areas

1. Process Listing & Inspection

# Basic process listing
ps aux                           # All processes, detailed
ps -ef                           # Standard format
ps --forest                      # Process tree

# Useful ps patterns
ps aux --sort=-%mem | head       # Top memory consumers
ps aux --sort=-%cpu | head       # Top CPU consumers

# Find specific processes
pgrep -f "pattern"               # PID by pattern
pgrep -l "nginx"                 # PID and name
pidof nginx                      # PID of named process
ps aux | grep '[n]ginx'          # Grep trick

# Top/htop
top                              # Real-time monitor
htop                             # Interactive (better UI)

2. Signal Handling

# Common signals
# SIGHUP (1)   - Reload config
# SIGINT (2)   - Ctrl+C
# SIGKILL (9)  - Force kill
# SIGTERM (15) - Graceful stop
# SIGSTOP (19) - Pause
# SIGCONT (18) - Resume

# Sending signals
kill PID                         # SIGTERM (default)
kill -9 PID                      # SIGKILL (force)
kill -HUP PID                    # SIGHUP (reload)

# Kill patterns
killall nginx                    # Kill by name
pkill -f "pattern"               # Kill by pattern

# Signal handling in scripts
cleanup() {
    echo "Caught signal, cleaning up..."
    rm -f "$TEMP_FILE"
    exit 0
}
trap cleanup SIGINT SIGTERM SIGHUP

3. Job Control

# Background execution
command &                        # Run in background
nohup command &                  # Immune to hangup
nohup command > output.log 2>&1 &

# Job management
jobs                             # List jobs
fg                               # Bring to foreground
bg                               # Resume in background
disown                           # Remove from shell

# Wait for processes
wait                             # Wait for all
wait $PID                        # Wait for specific

# Production pattern
start_daemon() {
    nohup ./daemon.sh >> /var/log/daemon.log 2>&1 &
    echo "$!" > /var/run/daemon.pid
    disown
}

4. Cron Scheduling

# Cron expression format
# ┌─────── minute (0-59)
# │ ┌───── hour (0-23)
# │ │ ┌─── day of month (1-31)
# │ │ │ ┌─ month (1-12)
# │ │ │ │ ┌ day of week (0-6)
# * * * * * command

# Common patterns
0 * * * *     # Every hour
*/15 * * * *  # Every 15 minutes
0 0 * * *     # Daily at midnight
0 0 * * 0     # Weekly on Sunday

# Edit crontab
crontab -e                       # Edit
crontab -l                       # List

# Production cron template
SHELL=/bin/bash
PATH=/usr/local/bin:/usr/bin:/bin
MAILTO=admin@example.com

# With locking
0 2 * * * /usr/bin/flock -n /var/lock/backup.lock /opt/scripts/backup.sh >> /var/log/backup.log 2>&1

5. Systemd Timers

# Timer unit: /etc/systemd/system/backup.timer
[Unit]
Description=Daily backup timer

[Timer]
OnCalendar=*-*-* 02:00:00
Persistent=true

[Install]
WantedBy=timers.target

# Timer management
systemctl enable backup.timer
systemctl start backup.timer
systemctl list-timers

Signal Quick Reference

Signal	Number	Action	Use Case
SIGHUP	1	Terminate	Reload config
SIGINT	2	Terminate	Ctrl+C
SIGKILL	9	Terminate	Force kill
SIGTERM	15	Terminate	Graceful stop
SIGSTOP	19	Stop	Pause
SIGCONT	18	Continue	Resume

Safety Guidelines

safety_rules:
  - rule: "Use SIGTERM before SIGKILL"
    reason: "Allow graceful shutdown"

  - rule: "Never kill PID 1"
    reason: "PID 1 is init"

  - rule: "Verify PID before killing"
    example: "ps -p $PID before kill"

cron_safety:
  - "Use flock to prevent overlap"
  - "Log all output"
  - "Set MAILTO for errors"

Troubleshooting Guide

Debug Checklist

1. ☐ Process exists: ps -p PID
2. ☐ Process owner: ps -o user= -p PID
3. ☐ Cron running: check /var/log/syslog
4. ☐ Environment: cron runs with minimal ENV

Common Issues Decision Tree

Process won't die?
├── Try: SIGTERM first, wait 30s
├── Check: if zombie (defunct)
├── Escalate: SIGKILL if stuck
└── Debug: strace -p PID

Cron job not running?
├── Check: crontab -l
├── Check: /var/log/syslog
├── Check: PATH and environment
└── Test: run script manually

References

• GNU Coreutils - Process
• Crontab Guru
• systemd Timers