Linux VPS

VPS disk still full after deleting files? Check these first

Trace missing disk space through filesystem checks, open deleted files and bounded log retention before deleting more data.

Close-up of an opened hard drive with its platter and read head visible
Photo: Borja Verbena / Unsplash. Display crop.

You delete a large archive, refresh your disk graph and still see 100% usage. Before removing anything else, check which filesystem is full and whether a running process still holds the deleted file open. A second round of random deletions can lose useful data without solving the cause.

These checks suit a Linux VPS that you administer. Read the output before making changes. If you only have a game-panel account, send the results you can access to support rather than trying host-level commands.

Compare the same filesystem

df -hT /
df -ih /
sudo du -xhd1 /

The first command reports allocated filesystem space; the second checks inode consumption. The last command adds up reachable files beneath the root directory, staying on that filesystem. Read the df and du manuals for the option details.

If your application lives on another mounted disk, repeat the checks against its mount point. Comparing root filesystem usage with a directory on a separate volume gives misleading totals. Run the directory scan during a quiet period on a busy server; reading millions of filenames can add I/O load.

Check errors from the scan too. An unprivileged account may miss directories it cannot read. Filesystem metadata, reserved space and snapshots can also explain differences. On snapshot-capable storage, inspect retention with the relevant filesystem tools before changing it.

Look for deleted files that remain open

sudo lsof +L1

The lsof manual documents this selection for files with fewer than one link. Match the command, process ID, file size and name to the application you expected. A deleted log can continue occupying disk blocks while its writer keeps the file descriptor open.

Save the evidence, then use that application's supported log-reopen or rotation procedure. Some services reopen logs on a documented signal; others need a planned restart. Check the service documentation instead of sending a signal to an unfamiliar process. Schedule a restart if players or database users will lose connections.

Repeat both the filesystem check and the open-file check after the change. Avoid truncating files through process-descriptor paths as a general fix. You could damage an active database or remove the evidence you need to understand recurring growth.

Check logs without deleting the log directory

sudo journalctl --disk-usage
sudo journalctl -p warning --since '2 hours ago'

A repeated application error can produce more logs than a retention policy expected. Find and correct that error before shrinking history. Keep incident logs you need for an investigation or support case on a separate volume with enough free space.

After reviewing retention requirements, an administrator might choose a bounded cleanup such as sudo journalctl --rotate --vacuum-size=500M. This example removes archived journal files and loses their history. It does not target application databases or ordinary text logs. The journalctl manual explains that vacuuming affects archived journals, while disk-usage totals include active files.

Check the application's own storage

Game worlds, database binary logs, temporary exports and backups can each need a separate retention policy. Ask the application which files it owns before removing them. A compressed backup stored beside its source may require room for both the archive and temporary working files.

For Docker workloads, inspect mounts and storage usage through Docker. Do not delete files inside its managed storage directory. Volume contents may hold the only copy of a world or database. A directory that looks like cache from the host can contain persistent application data.

Record what recovered the space

Write down the filesystem, free space before and after, the owning process and the action you took. Watch the next backup and the next busy session. If usage grows at the same rate, you have recovered breathing room rather than fixed the source.

Use our storage growth and backup retention guide to budget the next month, then add a free-space alert alongside the checks in what to monitor on a Linux VPS. Choose an alert threshold that leaves time for a safe response.

Sources and further reading