Common command-line debugging tools

df (disk free) shows how much space is used and available on each mounted filesystem — the first thing worth checking when something fails with a vague "no space left on device" error:

du (disk usage) shows what's actually consuming space within a directory — the natural next step once df confirms a disk is nearly full, narrowing down which specific directory is the actual culprit:

free shows memory usage — how much is used, available, and (on Linux) held in cache that can be reclaimed if something else needs it:

A system that's technically "out of free memory" isn't necessarily in trouble — much of "used" memory is often reclaimable cache, which is exactly why the available column (accounting for that reclaimable portion) is the more meaningful number to actually watch, not raw free.

On systems using systemd (most modern Linux distributions), journalctl shows system and service logs — including exactly what a service logged right before it crashed or failed to start:

For an application's own log files (not managed by systemd), the tools from earlier in this domain apply directly: tail -f app.log to watch it live, grep "ERROR" app.log to search it, combined together (tail -f app.log | grep "ERROR") to watch only the error lines as they happen in real time.

systemctl checks and controls services (long-running background programs managed by the system, like a web server or database):

systemctl status is very often the single fastest first check on an unresponsive server — it directly answers "is the actual service even running at all," before investigating anything more specific.

Putting this whole domain together, a genuinely common real scenario — "the app on this server stopped responding":

Every lesson here builds toward exactly this kind of moment: navigating to where the problem actually is, reading and searching the right files, checking permissions and ownership when something is unexpectedly inaccessible, understanding what's running and why, and having the specific command that answers the specific question a real failure raises — turning "something's wrong with the server" from a source of dread into a routine, answerable diagnosis.