mpstat Command
Display CPU statistics and performance metrics for multiprocessor systems.
Syntax
mpstat [OPTIONS] [INTERVAL [COUNT]]
mpstat [OPTIONS] -P {cpu_list|ALL} [INTERVAL [COUNT]]
The mpstat command displays activities for each available processor, processor 0 being the first one. Global average activities among all processors are also reported.
Common Options
| Option | Description |
|---|---|
-A |
Display all statistics (equivalent to -u -I ALL) |
-I {SUM|CPU|SCPU|ALL} |
Display interrupt statistics |
-P {cpu_list|ALL} |
Display statistics for specific processors |
-u |
Display CPU utilization report (default) |
-V |
Print version number and exit |
-o JSON |
Display statistics in JSON format |
-T |
Display timestamp for each report |
-h |
Display help message |
CPU Metrics Explained
| Metric | Description | Good Range |
|---|---|---|
%usr |
Percentage of CPU used by user processes | < 70% |
%nice |
Percentage of CPU used by nice processes | Variable |
%sys |
Percentage of CPU used by system processes | < 30% |
%iowait |
Percentage of CPU waiting for I/O operations | < 10% |
%irq |
Percentage of CPU servicing hardware interrupts | < 5% |
%soft |
Percentage of CPU servicing software interrupts | < 5% |
%steal |
Percentage of CPU stolen by hypervisor | < 5% |
%guest |
Percentage of CPU used by guest OS | Variable |
%idle |
Percentage of CPU idle time | > 20% |
Basic Examples
Basic CPU statistics
# Display current CPU statistics
mpstat
# Display statistics every 2 seconds
mpstat 2
# Display statistics every 5 seconds, 10 times
mpstat 5 10
# Display statistics with timestamp
mpstat -T 1
Basic CPU utilization monitoring
Per-CPU statistics
# Display statistics for all CPUs
mpstat -P ALL
# Display statistics for specific CPU
mpstat -P 0
# Display statistics for multiple CPUs
mpstat -P 0,1,2
# Monitor all CPUs every second
mpstat -P ALL 1
Monitor individual CPU cores and overall system
Interrupt statistics
# Display interrupt statistics
mpstat -I SUM
# Display per-CPU interrupt statistics
mpstat -I CPU
# Display software interrupt statistics
mpstat -I SCPU
# Display all interrupt statistics
mpstat -I ALL
Monitor interrupt activity and distribution
Advanced Usage
Comprehensive monitoring
# Display all available statistics
mpstat -A
# All statistics with per-CPU breakdown
mpstat -A -P ALL
# Comprehensive monitoring every 5 seconds
mpstat -A -P ALL 5
# All stats with timestamps
mpstat -A -T 1
Complete system monitoring with all metrics
Output formatting
# JSON output format
mpstat -o JSON 1 3
# JSON with all CPUs
mpstat -o JSON -P ALL 1 3
# Save output to file
mpstat 1 60 > cpu_stats.txt
# Append timestamp and save
mpstat -T 1 60 >> cpu_monitoring.log
Format output for analysis and logging
Specific monitoring scenarios
# Monitor high I/O wait systems
mpstat 1 | grep -v "Average"
# Monitor specific CPU under load
mpstat -P 0 1 | awk '$3 > 80 {print}'
# Monitor interrupt distribution
mpstat -I CPU -P ALL 1
# Monitor virtualized environment
mpstat 1 | awk '{print $1, $12, $13}' # Focus on steal and guest
Targeted monitoring for specific performance issues
Performance Analysis
Identifying CPU bottlenecks
# High CPU utilization check
mpstat 1 5 | awk 'NR>3 {if($4+$5 > 80) print "High CPU:", $0}'
# I/O wait analysis
mpstat 1 10 | awk 'NR>3 {if($7 > 10) print "High I/O wait:", $0}'
# System vs user CPU usage
mpstat 1 5 | awk 'NR>3 {print "User:", $4, "System:", $6}'
# Interrupt load analysis
mpstat -I SUM 1 5
Analyze CPU performance patterns and bottlenecks
Load balancing analysis
# Check CPU load distribution
mpstat -P ALL 1 5
# Identify unbalanced CPUs
mpstat -P ALL 1 | awk '$3 != "all" && $4 > 90 {print "Overloaded CPU:", $3}'
# Monitor interrupt distribution across CPUs
mpstat -I CPU -P ALL 1
# Check for CPU affinity issues
mpstat -P ALL 2 10 | grep -E "(CPU|all)"
Analyze workload distribution across CPU cores
Practical Examples
System monitoring scripts
# Basic monitoring script
#!/bin/bash
echo "CPU Monitoring Report - $(date)"
echo "================================"
mpstat 1 5 | tail -1
# Alert on high CPU usage
#!/bin/bash
CPU_USAGE=$(mpstat 1 1 | awk 'END{print 100-$NF}')
if (( $(echo "$CPU_USAGE > 90" | bc -l) )); then
echo "ALERT: High CPU usage: ${CPU_USAGE}%"
fi
# Log CPU stats to file
mpstat -T 60 >> /var/log/cpu_stats.log &
Automated CPU monitoring and alerting
Performance troubleshooting
# Diagnose performance issues
echo "=== CPU Overview ==="
mpstat
echo "=== Per-CPU Breakdown ==="
mpstat -P ALL
echo "=== Interrupt Analysis ==="
mpstat -I SUM
echo "=== 30-second monitoring ==="
mpstat 1 30 | tail -5
Comprehensive performance analysis workflow
Capacity planning
# Collect baseline metrics
mpstat 300 288 > daily_cpu_baseline.txt # 24 hours, 5-min intervals
# Peak usage analysis
mpstat 1 3600 | awk 'NR>3 {sum+=(100-$NF)} END {print "Avg CPU:", sum/(NR-3)}'
# Generate hourly summary
for hour in {0..23}; do
echo "Hour $hour:"
mpstat 3600 1 | tail -1
done
Long-term monitoring for capacity planning
Integration with Other Tools
Combining with system tools
# CPU and memory together
echo "CPU Stats:" && mpstat 1 1 | tail -1
echo "Memory Stats:" && free -h
# CPU with process information
mpstat 1 1 | tail -1 && echo "Top CPU processes:" && ps aux --sort=-%cpu | head -5
# CPU with load average
mpstat 1 1 | tail -1 && uptime
# CPU with I/O statistics
mpstat 1 1 | tail -1 && iostat -x 1 1 | tail -5
Combine mpstat with other monitoring tools
Data collection and analysis
# Collect data for analysis
mpstat -o JSON 1 60 > cpu_data.json
# Parse JSON output
mpstat -o JSON 1 3 | jq '.sysstat.hosts[0].statistics[].cpu-load[]'
# Create CSV for spreadsheet analysis
mpstat 1 60 | awk 'NR>3 {print $1","$4","$6","$7","$NF}' > cpu_stats.csv
# Real-time dashboard data
while true; do
mpstat 1 1 | tail -1 | awk '{print $4,$6,$7,$NF}'
sleep 1
done
Collect and format data for further analysis
Troubleshooting
Common Issues
- Command not found - sysstat package not installed
- No data displayed - Insufficient permissions or system issues
- Inconsistent readings - System under heavy load
- Missing CPU cores - CPU hotplug or virtualization issues
Installation and setup
# Install sysstat package
# Ubuntu/Debian
sudo apt update && sudo apt install sysstat
# RHEL/CentOS/Fedora
sudo yum install sysstat
# or
sudo dnf install sysstat
# Verify installation
mpstat -V
# Check if service is running
systemctl status sysstat
Install and configure sysstat package
Common troubleshooting
# Check system load
uptime
# Verify CPU information
lscpu
cat /proc/cpuinfo | grep processor | wc -l
# Check for CPU throttling
cat /proc/cpuinfo | grep MHz
# Monitor system messages
dmesg | grep -i cpu
# Check process consuming CPU
top -o %CPU
Diagnose system and CPU-related issues
Best Practices
Monitoring Guidelines
- Use appropriate intervals - 1-5 seconds for troubleshooting, 5-15 minutes for trending
- Monitor all CPUs on multi-core systems with -P ALL
- Include timestamps for historical analysis
- Combine with other tools for complete system view
- Set up automated alerts for critical thresholds
- Archive historical data for capacity planning
Performance Interpretation
- High %usr - Application bottleneck, consider optimization
- High %sys - System call overhead, kernel bottleneck
- High %iowait - I/O bottleneck, check disk performance
- High %steal - Virtualization overhead, resource contention
- Uneven CPU distribution - Load balancing issues
- High interrupt rates - Hardware or driver issues