ENOSPC: no space left on device — how to fix it
Quick answer: ENOSPC: no space left on device means a write failed because something is full — usually the disk, but it can also be inodes or the inotify watch limit. Check disk with df -h, inodes with df -i, and for the dev error "ENOSPC: System limit for number of file watchers reached" raise fs.inotify.max_user_watches. Then reclaim space — logs and old Docker images are the usual offenders.
What the error looks like
Same error code, three different "full" conditions:
Error: ENOSPC: no space left on device, write
# the dev-server flavor:
Error: ENOSPC: System limit for number of file watchers reached
# docker:
failed to register layer: no space left on deviceDon't assume "disk full" — a disk with free space can still throw ENOSPC if it's out of inodes or if a watcher hit the inotify limit. Check all three.
Why it happens
The disk is genuinely full
Logs, build artifacts, or a runaway file filled the partition. df -h shows 100%.
Inodes are exhausted
Millions of tiny files used up all inodes while bytes remain. df -i shows 100% IUse.
Docker images and volumes piled up
Old images, dangling layers, and stopped containers quietly consume the disk.
inotify watch limit (dev)
A dev server watching many files hit max_user_watches — not a disk problem at all.
Diagnose it in three steps
Check space and inodes
df -h # bytes used
df -i # inodes used (a sneaky cause)Find what's eating it
du -sh /var/* 2>/dev/null | sort -h | tail
journalctl --disk-usage
docker system dfReclaim or raise the watch limit
# disk:
docker system prune -af --volumes
journalctl --vacuum-size=200M
# dev file-watcher limit:
sudo sysctl fs.inotify.max_user_watches=524288Stop the disk filling in the first place
Reclaiming space gets you running again; the durable fix is bounding what grows. Rotate logs, prune images on a schedule, and put high-churn data on a volume you monitor:
# rotate app/journal logs so they can't fill the root disk
# (logrotate, journald SystemMaxUse, or your platform's log limits)
# and alert before df hits ~85%, not at 100%.Logs that won't fill your disk
Runaway logs are a top cause of a full disk — and a control plane that streams logs has to handle that without taking the host down. Infraveil runs on your own servers and is built so log collection won't fill your disk, with service status and disk-related failures surfaced in one place instead of discovered when a write fails.
Frequently asked questions
What does ENOSPC mean?
A write failed because a resource is full. Most often the disk, but it can also be inode exhaustion or, in dev, the inotify file-watch limit.
My disk has space but I still get ENOSPC — why?
You're likely out of inodes (check df -i) from many tiny files, or you hit the inotify max_user_watches limit if it's a dev server.
How do I fix 'System limit for number of file watchers reached'?
Raise the limit: sudo sysctl fs.inotify.max_user_watches=524288 and persist it in /etc/sysctl.conf. It's a watcher limit, not a disk issue.
How do I reclaim Docker disk space?
Run docker system df to see usage, then docker system prune -af --volumes to remove unused images, containers, and volumes (be careful with --volumes).