AllMC 1.7.xMC 1.12.xMC 1.16.xMC 1.19.x

No space left on device (Minecraft IOException)

'java.io.IOException: No space left on device' means the disk holding your Minecraft world is full. The JVM can't flush the world to disk, can't write a crash report, and can't even rotate the log file — so the server hard-crashes the moment it tries any IO. The fix is finding what's eating the space and reclaiming it.

Fastest path: paste your crash report into the Crash Doctor — it identifies this pattern and 40+ others automatically. No signup.

How to fix it

  1. 1

    Identify the largest space hogs

    In the Files tab, open the server root and sort by size. The usual suspects: backups/ (auto-backups accumulate), world*/ (especially nether and end dimensions on long-running worlds), logs/ (verbose mods can write hundreds of MB per day), and crash-reports/ if the server has been crash-looping.

  2. 2

    Delete old backups first

    Backups are the easiest reclaim — they're not load-bearing once a known-good backup is offsite. Keep the last 3-5; delete the rest. If you host with CoalHosting our backup retention is configurable per server in Settings → Backups.

  3. 3

    Prune unvisited dimensions

    If players never go to the nether or the end, those dimension folders are wasted disk. Stop the server, move world/dimensions/the_nether and the_end out, then start. Worst case players regenerate them on first visit; best case the disk stays freed forever.

  4. 4

    Upgrade the plan if the world is genuinely big

    Hosted plans cap disk per RAM tier (the same VM hosts both). If you're running a 5-year-old world with 50k explored chunks, you may have outgrown the plan. The fix is more disk — usually a RAM bump on hosted services because the two are tied.

Alternative causes

These can produce the same error message — worth ruling out if the steps above don't resolve it.

Inode exhaustion (filesystem out of file slots, not bytes)

Rare but possible: a mod that writes millions of tiny files can exhaust inodes before bytes. df -i shows inode usage. Symptom: 'No space' even though df shows free GB. Fix is the same — delete or consolidate the file flood.

Quota set lower than the visible disk

On hosted environments the underlying disk may be huge, but the per-server quota is what the JVM sees. If the quota is full but the physical disk isn't, contact your host — they'll see the quota hit and can raise it.

Frequently asked

Will deleting the world folder break the server?

Yes — that's the live world. Don't touch world/ itself. Backup/ and old-world/ folders are safe to delete. Dimension subfolders inside world/ (world/the_nether, world/the_end) regenerate if removed.

How big does a Minecraft world get over time?

Roughly 1-2 MB per fully explored chunk. A small SMP with 5k explored chunks: ~10 GB. A long-running 30+ player server: 50-200 GB+. Modpacks with custom dimensions or worldgen mods (Twilight Forest, Aether) inflate this significantly.

Why does the server crash AFTER deleting files?

Sometimes — file deletion frees space lazily on some filesystems. If you delete 10 GB and restart immediately, the JVM may still see the disk as full. Stop the server, wait 30 s, start again.

Want this auto-fixed on your server?

CoalHosting's Minecraft hosting runs the same pattern database against every crash and applies the known fix before your players notice. Free crash diagnosis works on any server, hosted or not.

Related crashes

Last reviewed 2026-06-05. If a step is wrong or out of date, tell us — we'll fix the article and the auto-pattern at the same time.