Linux Survival Basics

Watch Logs Without Opening the Whole File

You need to watch recent log lines while a service or script is actively failing.

Command

tail -n 80 -f /var/log/nginx/error.log

What changed

Nothing changes. The command streams the end of the file so new failures are visible immediately.

Danger

safe

When to use it

Use this while reproducing a web server, script, cron, or service failure.

When not to use it

Do not leave noisy log tails running forever on a shared terminal.

Undo or recovery

Press Ctrl-C to stop following the file.

Expected output

Recent log lines followed by new lines as they are written.

demo script

Disposable terminal steps

tail -n 20 /var/log/nginx/error.log
tail -n 80 -f /var/log/nginx/error.log
printf 'simulate a new error line\n' >> /var/log/nginx/error.log

Terminal transcript Demo JSON

simulated output

What it looks like

disposable vessel

::fixture-ready::
$ tail -n 20 /var/log/nginx/error.log
2026/06/25 10:00:01 [notice] 11#11: start worker process
2026/06/25 10:01:42 [error] 11#11: *7 connect() failed (111: Connection refused) while connecting to upstream
2026/06/25 10:02:05 [warn] 11#11: upstream server temporarily disabled
2026/06/25 10:03:16 [error] 11#11: *9 access forbidden by rule, client: 203.0.113.44
2026/06/25 10:04:33 [error] 11#11: *10 upstream timed out while reading response header
::exit-code::0
$ tail -n 80 -f /var/log/nginx/error.log
2026/06/25 10:00:01 [notice] 11#11: start worker process
2026/06/25 10:01:42 [error] 11#11: *7 connect() failed (111: Connection refused) while connecting to upstream
2026/06/25 10:02:05 [warn] 11#11: upstream server temporarily disabled
2026/06/25 10:03:16 [error] 11#11: *9 access forbidden by rule, client: 203.0.113.44
2026/06/25 10:04:33 [error] 11#11: *10 upstream timed out while reading response header
::exit-code::124
$ printf 'simulate a new error line\n' >> /var/log/nginx/error.log
::exit-code::0

YouTube Short

Watch the failure happen live.

Do not open the whole log file. Tail the last useful lines, reproduce the issue, and watch the error arrive.

experiments

A/B tests to run

Metric: youtube_retention_15s

A: Watch the failure happen live.

B: Stop opening giant log files.