VPS Disk Full: How to Find and Clean Up Space

A 100% full disk is worse than it sounds: the database refuses writes, the site throws 500s, and even logs stop recording — leaving you blind. Here is a standard procedure from locating to cleaning.

Step 1: assess globally

Run df -h to see usage per partition and confirm whether the root partition or some mount point is full. If inodes are exhausted (df -i shows 100%) while space remains, you have a tiny-files problem — same approach, but hunt for directories with huge file counts.

Step 2: locate big directories

Drill down from root: du -h --max-depth=1 / 2>/dev/null | sort -rh | head. Enter the largest directory and repeat — two or three rounds pinpoint the culprit.

The five usual suspects

• Logs: rotated logs under /var/log and logs written by your own apps. Shrink the system journal with journalctl --vacuum-size=200M and configure logrotate for app logs;
• Docker: dangling images and build cache consume huge space — inspect with docker system df, clean with docker system prune (confirm nothing in use gets removed);
• Old backups: backup scripts that only add and never delete fill a disk within months — add a retention policy;
• Package caches: apt clean frees download caches;
• Deleted-but-open files: space is not freed while a process still holds a deleted file's handle — find it with lsof | grep deleted and restart that process.

Two iron rules when cleaning

First, look before you delete: confirm what a file is for — never rm database directories or actively-written logs. Second, truncate large active files instead of deleting: emptying a live log with truncate -s 0 is safer than removing it. Afterwards, set a disk usage alert (warn above 80%) so the next incident never reaches 100%.