CI /tmp Directory Full (ENOSPC on tmpfs)

By Kaveh Alemi·Latchkey

A write to /tmp failed with ENOSPC. Either the temp directory accumulated files or /tmp is a small tmpfs (RAM-backed) mount that filled quickly.

What this error means

A step fails writing to /tmp with No space left on device, often during compression, test fixtures, or a downloaded installer. Clearing /tmp or redirecting TMPDIR fixes it with no code change.

gzip: stdout: No space left on device
mktemp: failed to create file via template '/tmp/build.XXXXXX': No space left on device

Common causes

Temp files accumulated across steps

Tools leave behind temp files; on a reused runner /tmp can fill before a job cleans up.

/tmp is a small tmpfs

When /tmp is RAM-backed and sized small, a large temp write exhausts it even with disk free.

How to fix it

Clear /tmp or point TMPDIR at a bigger path

Reclaim temp space, or redirect temporary writes to the work disk.

df -h /tmp
rm -rf /tmp/* 2>/dev/null
export TMPDIR="$RUNNER_TEMP"

Reduce temp usage per step

1. Stream large data instead of staging it in /tmp.
2. Clean up temp files within the step that creates them.
3. Use a runner with a larger disk for temp-heavy work.

How to prevent it

• Set TMPDIR to the work volume for temp-heavy jobs.
• Clean temp files within each step.
• Right-size disk and tmpfs for the workload.

Frequently asked questions

Related guides

References

• GitHub Actions documentation