How to check free disk space on a Linux server
df -h is the free-disk-space reading on a Linux server. Use the row whose Mounted on column is /, or the mount that holds the site. Avail is the space you can still write. Use% near 100 means the disk is full.
df reports filesystems and du reports folders. On a Linux VPS you run both over SSH on the guest. A desktop disk tool is not on a headless server, and it would not see this mount.
Use these four commands, in this order, before you delete anything.
| Command | What it reads | When to run it |
|---|---|---|
| df -h | Size, Used, Avail, Use% and the mount | First look at free disk space |
| df -i | Inode use, not bytes | Writes fail but Avail is not zero |
| du -xh --max-depth=1 / | One level of /, same filesystem | Find which top folder is heavy |
| du -sh /var/log | Size of one path you name | Confirm a folder before you delete |
Ignore tmpfs and udev rows. They live in memory, not on the disk in the plan. If several disks are mounted, clean the one with the high Use%. df -h --total adds a total line that mixes those memory mounts in, so trust the / row.
ls -lh and stat report one file. fdisk -l reports partition size. None of those is free space. The free-space number is Avail from df -h.
What fills a Linux server disk: logs, backups and cache
Logs, on-disk backups and package cache fill a Linux server disk, in that order. Databases grow too, but you measure them and you do not delete their files.
Start one level down from / and stay on that filesystem, so du does not walk a second disk.
du -xh --max-depth=1 / prints that level. Open the largest name next, usually /var, /home, /usr or /opt, and run du again there.
- /var/log and the journal — web, mail and database logs, plus the systemd journal, grow on every busy day.
- Backups on the same disk — a plugin that keeps every copy under the site, or a dump left in /root, uses the disk the site needs.
- Package cache — apt, yum and dnf keep downloaded packages under /var/cache after the software is already installed.
- Databases — du -sh /var/lib/mysql or /var/lib/postgresql shows the data directory. Do not delete ibdata, .ibd files or the PostgreSQL data directory.
A size query lists schemas and does not free space: SELECT table_schema, ROUND(SUM(data_length + index_length)/1024/1024, 1) AS mb FROM information_schema.tables GROUP BY table_schema ORDER BY mb DESC; Shrink a database with a planned dump, not with rm.
Keep one copy of any backup off this server. Server backup is that copy. A file that exists only on this disk is not a backup you can delete.
Safe commands to free disk space on a Linux server
Delete only after df and du agree on the mount and the folder. Run df -h after each command. If Avail does not rise, a process may still hold the file.
- Cap the systemd journal with journalctl --vacuum-size=100M. It prints how much it freed. It does not touch the text files in /var/log.
- Remove compressed logs older than 30 days with find /var/log -type f -name '*.gz' -mtime +30 -delete. Leave the live syslog, auth log and mail log in place.
- Clear downloaded packages, not installed software. On Debian and Ubuntu run apt-get clean. On the RHEL family run yum clean all or dnf clean all.
- Remove temp files untouched for a week with find /tmp -type f -atime +7 -delete. Skip this while a deploy or an installer is running.
- List your own old backups by name, then rm those paths. Never rm a wildcard across /var, /home or the site root.
If you removed a large file and df -h did not move, a process still has it open. lsof +L1 lists deleted files that are held open. Restart that service, then read Avail again.
Do not delete /var/lib/mysql, a PostgreSQL data directory, /var/spool, /etc or the document root to make space. Those paths are the site.
How logrotate keeps a Linux server disk free
logrotate renames, compresses and drops old logs so /var/log cannot grow without a cap. The distro already runs it from /etc/logrotate.conf and /etc/logrotate.d/.
A safe web-log block rotates weekly, keeps four compressed copies, and skips empty or missing files. Dry-run a new file first: logrotate -d /etc/logrotate.conf. The -d flag prints the plan and changes nothing.
Do not rm a log that a daemon still has open. logrotate can copy and truncate, or it can signal the daemon to reopen the file. Deleting the live file leaves the daemon writing to a removed inode, and df stays full.
When inodes make a Linux server look full
An inode is one file or directory. The disk can show free gigabytes and still refuse writes when the inode table is full. df -h looks fine. df -i shows IUse% at 100.
Tiny files cause it: sessions, mail, cache objects, and a backup plugin that never deletes old copies. One large database file does not. On GNU du, du --inodes -xd 1 / shows which directory holds the count.
Delete the old small files by age and by name, then run df -i again. A bigger disk does not help if you keep creating millions of files on it.
How to stop a Linux VPS disk filling up again
A Linux server disk is a fixed size. Unlimited disk space is a shared-hosting slogan, not a property of this mount. Clean when Use% on / passes 80, before MySQL or mail stops.
A weekly SSH check of df -h / and df -i / is enough. Bandwidth monitoring measures transfer, not bytes stored on disk, so it will not warn you that / is full.
If Avail is still near zero after the safe deletes, the workload needs a larger disk on the VPS plan.
If you cannot get a shell because the disk is full, email support@webji.services. That mailbox is staffed 24/7 in Hindi and English. The office phone is not a support line.
Frequently Asked Questions
How do I free disk space on a Linux server?
Run df -h, then du -xh --max-depth=1 / on the full mount. Delete rotated logs, package cache and backups copied off the server, then run df -h again. Stop if Avail does not grow.
How do I check free disk space on Linux?
df -h prints Size, Used, Avail and Use% for each mount. Read the row for /. Run df -i when the error is No space left on device but Avail is not zero.
What is eating my server's disk space?
Usually /var/log, a backup folder on the same disk, or /var/cache. du -xh --max-depth=1 / names the heavy top-level directory. Databases show up under /var/lib/mysql or /var/lib/postgresql.
How do I clean /var/log safely?
Cap the journal with journalctl --vacuum-size=100M, and delete only compressed logs older than 30 days with find /var/log -type f -name '*.gz' -mtime +30 -delete. Leave the live log files for logrotate.
What are inodes and can they run out?
Yes. Each file uses one inode, and df -i shows the count. A backup plugin or a session folder can hit 100 percent IUse while df -h still shows free gigabytes. Remove those small files, not the database.
How do I find the largest files on Linux?
find / -xdev -type f -size +200M -exec du -h {} + | sort -h lists big files on this filesystem only. Confirm each path, then delete a named file. Do not remove a database file because it is large.
Why does my VPS disk keep filling up?
Logs and on-server backups grow back when nothing rotates them. logrotate plus a weekly df -h / check stops the repeat. The disk is a fixed size, so a workload that stays full after cleanup needs a larger plan.