Linux Disk Space Is Full: Diagnose Before Deleting Files

A full filesystem can break database writes, sessions, logs, updates, and even SSH operations. The first task is to identify which filesystem is full and whether the issue is capacity or inode exhaustion.

Use df -h to inspect space and df -i to inspect inode usage. Then identify large directories with du, starting from the affected mount. Deleted files may still consume space when a process has them open; lsof can help locate those handles. Do not delete unfamiliar database files or active service files to recover space quickly.

Logs, backups, cache directories, container layers, and old releases are common sources of growth. Establish which files are safe to rotate or remove, and preserve at least one known-good backup outside the filesystem being cleaned. Check retention policies before deleting backups that may be the only recovery point.

After cleanup, verify the service that failed and add monitoring or retention limits. A recurring disk alert is easier to handle than an emergency cleanup while production writes are failing.