GitHub Actions "Failed to restore: Cache service responded with 503"

By Daniel Zoghalchali·Latchkey

The Actions cache service returned HTTP 503 (Service Unavailable) on a restore. This is a temporary backend outage - the cache is skipped, the job continues, and a retry typically restores normally.

What this error means

An actions/cache restore step logs "Warning: Failed to restore: Cache service responded with 503". It is a warning, not a failure, so the job proceeds with a cache miss.

Warning: Failed to restore: Cache service responded with 503

Common causes

Transient cache backend unavailability

A brief outage or degraded window in the cache service returns 503 to restore requests until it recovers.

Restore treated as fatal

If a workflow wraps the cache step to fail on any error, a transient 503 wrongly fails the job instead of falling back to a clean install.

How to fix it

Let the job continue on a restore miss

Keep the cache step non-fatal so a 503 falls back to installing dependencies normally.

- uses: actions/cache@v4
  id: cache
  continue-on-error: true
  with:
    path: ~/.cache/pip
    key: pip-${{ hashFiles('requirements.txt') }}
- run: pip install -r requirements.txt

Retry the transient outage

1. Re-run the job; 503s from the cache service are short-lived.
2. Do not gate the build on a successful cache restore.
3. Check the GitHub status page if 503s persist across many runs.

How to prevent it

• Treat cache restore as best-effort, never required.
• Always keep the real install step so a miss is harmless.
• Retry transient 503s instead of failing the job.

Frequently asked questions

Related guides

References

• GitHub Actions documentation
• Workflow syntax for GitHub Actions