Disk Full on a Linux Server: How to Find What's Eating Space

A server throws No space left on device and suddenly things break — apps can't write, logs stop, deploys fail. A full disk is an emergency, but tracking down the cause is quick once you know the two or three commands that matter.

Step 1: Confirm the disk is actually full

df -h

This lists each filesystem with its size and how full it is. Find the one at or near 100% — that's the one you need to clear. (df reporting full while files seem small can also mean inodes are exhausted, but a genuinely full disk is the common case.)

Step 2: Find the biggest directories

Starting from the full filesystem (often /), find what's hogging space:

du -h --max-depth=1 / | sort -h

That shows the size of each top-level directory, largest last. Follow the trail downward into the biggest one, repeating the command, until you find the actual culprit.

The usual suspects

• Log files — runaway logs in /var/log are the classic cause, sometimes a single huge file.
• Old container images and volumes — Docker can quietly eat enormous space over time.
• Build artifacts, caches, and old releases piling up from repeated deploys.
• Forgotten backups or dumps left in a home directory.

Step 3: Reclaim space safely

Delete with care — confirm what something is before removing it. Truncating or rotating a giant log is usually safe; if Docker is the offender, its built-in cleanup reclaims unused data:

docker system df       # see what Docker is using
docker system prune    # remove unused data (read the prompt first)

Never blindly delete files you don't recognise to free space — you can break the system worse than a full disk. Identify first, then remove.

The takeaway

df -h to find the full filesystem, du to walk down to the biggest offender, then clear it deliberately. Logs and Docker leftovers are the usual villains, and a few minutes with these commands turns a scary outage into a routine cleanup.