mpstat

Understanding mpstat’s Output

mpstat’s output can seem daunting at first, but with a little explanation, it becomes readily understandable. By default, mpstat displays average CPU utilization statistics since the system booted. However, its real power lies in its ability to provide real-time updates and historical data.

The key metrics you’ll see include:

• %user: Percentage of CPU time spent running user-level processes. High values might indicate a workload-intensive application or a process hogging resources.
• %nice: Percentage of CPU time spent running user-level processes with niced priority (lower priority).
• %system: Percentage of CPU time spent running kernel-level processes. High values could suggest kernel issues or driver problems.
• %iowait: Percentage of CPU time spent waiting for I/O operations. This is a strong indicator of disk bottlenecks.
• %irq: Percentage of CPU time spent servicing hardware interrupts.
• %softirq: Percentage of CPU time spent servicing software interrupts.
• %steal: Percentage of CPU time stolen by another virtual machine in a virtualized environment. Relevant only in virtualized setups.
• %idle: Percentage of CPU time spent idle.

Basic Usage: Getting a Snapshot of CPU Activity

The simplest way to use mpstat is to run it without any arguments:

mpstat

This will show the average CPU statistics since boot. To get a more detailed view, including per-core statistics, use the -P option:

mpstat -P ALL

-P ALL displays statistics for all CPUs and cores. You can specify a particular CPU core using a number, for example:

mpstat -P 0  # Statistics for CPU 0

Monitoring CPU Performance Over Time

For continuous monitoring, specify the interval and number of samples:

mpstat 2 5  # Display statistics every 2 seconds for 5 samples

This command will show CPU utilization every two seconds for five iterations. This is useful for observing CPU behavior during resource-intensive tasks.

Focusing on Specific Metrics

While the default output is detailed, you might only need certain metrics. Using the -u option displays only user and system statistics:

mpstat -u 2 5

Similarly, the -I option can be used to show interrupt statistics (IRQ and softirq):

mpstat -I SUM 2 5 # SUM gives aggregated interrupt stats

Analyzing the Results

By analyzing the output of mpstat, you can identify performance bottlenecks. For instance:

• Consistently high %iowait suggests disk I/O is a limiting factor.
• High %user combined with low %idle could indicate a CPU-bound process.
• Elevated %system might hint at kernel-level problems or inefficient drivers.

Understanding these relationships is key to using mpstat effectively for performance tuning and troubleshooting.

Advanced Usage: Averaging over Intervals

mpstat also offers the ability to average statistics over specific time intervals using the -A option:

mpstat -A 10 1 # Averages over 10 seconds for one sample

This provides a smoother view of CPU utilization compared to snapshots taken at short intervals.

Using these examples and understanding the core metrics, you can effectively use mpstat to gain a better understanding of your Linux system’s performance and optimize its resource usage. This detailed understanding will allow you to proactively address performance issues before they impact your system’s stability and responsiveness.