Docker /dev/shm No Space Left On Device

A container can fail with write /dev/shm/...: no space left on device while the host disk still has plenty of free space. Treat it as a shared-memory, session, or health-check loop incident before deleting app data or Docker volumes.

Docker deletion judgment

Find which Docker storage can be deleted.

Leave your email now; Docker or OrbStack storage numbers can follow after the first reply. We send the free SafeDisk AI deletion trial step only if the storage still needs review.

See sample result Ask AI about one file

First evidence

Separate host disk, container shared memory, and runaway sessions.

These checks are read-only. They show whether the failure is ordinary disk pressure, a tiny default 64 MB /dev/shm, or a request loop creating too many session artifacts.

container="YOUR_CONTAINER_NAME"

echo "== host and container filesystems =="
df -h /
docker exec "$container" sh -lc 'df -h / /tmp /dev/shm 2>/dev/null || true'
docker exec "$container" sh -lc 'df -i / /tmp /dev/shm 2>/dev/null || true'

echo "== largest /dev/shm entries =="
docker exec "$container" sh -lc 'find /dev/shm -maxdepth 2 -type f -printf "%s %p\n" 2>/dev/null | sort -nr | head -40'

echo "== recent shared-memory/session errors =="
docker logs --since=2h "$container" 2>&1 | grep -Ei "dev/shm|no space left|SIGBUS|session|health" | tail -80

Ask free AI judgment

What Not To Delete First

Do not prune Docker volumes because /dev/shm is full. Volumes may contain databases or app state and may not be related.
Do not delete random session files while active workers are still writing. Stop the loop or route first, then clean stale files.
Do not only increase shm_size if a health check or redirect loop is creating a new session every second.
Do not treat SIGBUS workers as a normal app crash until shared-memory pressure is measured.

Common Root Causes

Docker's default container /dev/shm is too small for the workload.
A reverse proxy or health check hits an authenticated route and creates session files repeatedly.
Browser automation, databases, Python workers, or image tooling use shared memory for temporary state.
Workers crash and restart while leaving stale shared-memory files behind.
The app logs the symptom as HTTP 400 or worker SIGBUS, hiding the storage boundary.

Copy-ready issue reply

Use this when maintainers ask for useful logs.

This keeps the thread focused on evidence they can act on without exposing secrets or deleting data.

I would capture this as a /dev/shm pressure incident, not ordinary host disk-full yet.

Useful evidence:
- container `df -h /dev/shm /tmp /` and `df -i /dev/shm /tmp /`
- largest files under `/dev/shm`, grouped by prefix such as session IDs
- request logs around the growth window, especially health checks, redirects, and unauthenticated paths
- worker restart logs showing SIGBUS or "no space left on device"
- compose/run config showing `shm_size` or Docker's default 64 MB shared memory

Acceptance check for a fix:
- the health-check route no longer creates sessions
- `/dev/shm` usage stabilizes over the previous failure window
- stale session cleanup only runs after active writers are stopped or aged out
- alerting watches `/dev/shm` free bytes separately from host disk free bytes

Ask free AI judgment

Still not sure what is safe?

Copy the read-only check first. Request the free SafeDisk AI deletion trial only if the output, app state, or cleanup boundary is still not obvious.

Ask free AI judgment See sample result

Free AI deletion trial

Need a delete / confirm / protect answer for Docker?

Submit the form first; the storage summary can follow. We check whether free guidance is enough before asking for the free SafeDisk AI deletion trial.