Master Linux Performance: 10 Essential Commands to Diagnose Issues in 60 Seconds

When you encounter performance problems on a Linux server, the first minute is crucial. This article from Netflix’s performance engineering team lists ten standard Linux command‑line tools you can run to get a quick overview of system health and process resource usage.

The approach follows the USE method (Utilization, Saturation, Errors) and focuses first on error and saturation metrics before looking at utilization.

uptime

uptime
 dmesg | tail
 vmstat 1
 mpstat -P ALL 1
 pidstat 1
 iostat -xz 1
 free -m
 sar -n DEV 1
 sar -n TCP,ETCP 1
 top

The uptime command shows the load average for the past 1, 5 and 15 minutes, indicating how many tasks are waiting to run.

dmesg | tail

Shows the most recent kernel messages, useful for spotting OOM‑killer events or TCP errors.

vmstat 1

Displays virtual memory and CPU statistics every second. Important fields include r (runnable tasks), free (free memory), us / sy (user and system CPU time), and wa (IO wait).

mpstat -P ALL 1

Shows per‑CPU utilization, helping identify uneven CPU usage or single‑threaded bottlenecks.

pidstat 1

Provides per‑process CPU usage over intervals, allowing you to spot processes that consume many CPU cores.

iostat -xz 1

Reports block device statistics such as r/s, w/s, await, avgqu‑sz and %util, which reveal disk saturation.

free -m

Shows memory usage, distinguishing between buffers, cache and actual used memory; the “‑/+ buffers/cache” line gives a more accurate view of free memory.

sar -n DEV 1

Monitors network interface throughput ( rxkB/s, txkB/s) and interface utilization.

sar -n TCP,ETCP 1

Provides TCP statistics such as active connections, passive connections and retransmissions, useful for assessing network health.

top

Combines many of the above metrics in a dynamic view; however, because it refreshes the screen, transient spikes may be missed, so tools like vmstat or pidstat are often preferable for continuous monitoring.

By running these commands within the first 60 seconds of investigation, you can quickly narrow down whether the problem lies in CPU saturation, memory pressure, disk I/O, or network issues.