How to Diagnose Linux Server Performance Issues in the First 60 Seconds

When a Linux server shows performance problems, the first minute is crucial. Netflix’s performance engineering team relies on a set of standard Linux command‑line tools to gather key metrics quickly.

The approach focuses on three aspects—Utilization, Saturation, and Error (USE)—to evaluate CPU, memory, disk, and network resources.

1. uptime

<code>$ uptime</code>

Shows the system’s average load over the last 1, 5, and 15 minutes, indicating how many tasks are waiting for CPU or I/O.

2. dmesg | tail

<code>$ dmesg | tail</code>

Displays the most recent kernel messages, helping to spot errors such as OOM kills or TCP issues.

3. vmstat 1

<code>$ vmstat 1</code>

Provides virtual memory and CPU statistics every second. Important fields include:

r : runnable tasks (CPU saturation indicator)

free : free memory in KB

si, so : pages swapped in/out (memory pressure)

us, sy, id, wa, st : CPU time spent in user, system, idle, I/O wait, and steal.

High

wa

suggests I/O bottlenecks; high

sy

(>20%) may indicate kernel overhead.

4. mpstat -P ALL 1

<code>$ mpstat -P ALL 1</code>

Shows per‑CPU utilization, useful for detecting uneven load or single‑threaded hotspots.

5. pidstat 1

<code>$ pidstat 1</code>

Similar to

top

but provides periodic snapshots of per‑process CPU usage, making it easier to track processes that consume many cores.

6. iostat -xz 1

<code>$ iostat -xz 1</code>

Reports block device statistics. Key metrics:

r/s, w/s, rkB/s, wkB/s : read/write request rates and throughput

await : average I/O response time (high values indicate saturation)

avgqu‑sz : average queue length (values >1 suggest bottleneck)

%util : device utilization (values >60% often cause performance issues).

7. free -m

<code>$ free -m</code>

Shows memory usage, distinguishing between used, free, buffers, and cache. The “-/+ buffers/cache” line gives a more accurate view of usable memory.

8. sar -n DEV 1

<code>$ sar -n DEV 1</code>

Monitors network interface throughput (rxkB/s, txkB/s) and utilization (%ifutil) to identify possible network bottlenecks.

9. sar -n TCP,ETCP 1

<code>$ sar -n TCP,ETCP 1</code>

Provides TCP statistics such as active connections per second, passive connections, and retransmissions, which help assess network health.

10. top

<code>$ top</code>

Aggregates many of the above metrics in real time, allowing a quick sanity check of system load and process activity. For detailed trend analysis, tools like

vmstat

or

pidstat

are preferred because they can log data over time.