Docker Compose Pull Fails with Rate Limit / 5xx - Fix Flaky Pulls

Q: How do I fix "compose pull" failed?

Log in, then pull images as a separate retryable step so up does not re-fetch.

By Kaveh Alemi·Latchkey

docker compose pull (and the implicit pull in up) fetches every service image. Any one of those pulls can hit a Hub rate limit or a registry 5xx - a transient failure, not a compose bug.

What this error means

docker compose up/pull fails while fetching one service’s image with toomanyrequests or received unexpected HTTP status: 5xx. Re-running often succeeds once the limit resets or the registry recovers.

docker compose output

db Pulling
db Error   toomanyrequests: You have reached your pull rate limit.
Error response from daemon: ... pull rate limit

Common causes

Anonymous pulls share a Hub rate limit

Compose pulls several base images; unauthenticated on a shared CI IP, the combined pulls hit Docker Hub’s anonymous rate limit quickly.

Registry-side 5xx during a pull

A transient registry outage can fail one image pull mid-up, aborting the whole compose start even though the file is fine.

How to fix it

Authenticate and pull before up

.github/workflows/test.yml

echo "${{ secrets.DOCKERHUB_TOKEN }}" | docker login -u "${{ secrets.DOCKERHUB_USERNAME }}" --password-stdin
for i in 1 2 3; do docker compose pull && break; sleep $((i*10)); done
docker compose up -d --no-build

Mirror or cache compose images

Pull service base images from a registry you control or a pull-through cache.
Pin image digests so cached layers are reused across runs.
Cache images between jobs to avoid re-pulling on every run.

How to prevent it

Authenticate to Docker Hub before any compose pull in CI.
Mirror compose service images into your own registry.
Wrap compose pull in a bounded retry with backoff.

Frequently asked questions

What causes ""compose pull" failed"?

Compose pulls several base images; unauthenticated on a shared CI IP, the combined pulls hit Docker Hub’s anonymous rate limit quickly.

How do I fix "compose pull" failed?

Can Latchkey fix this automatically?

Yes. Latchkey runs your GitHub Actions on managed runners that detect this failure, apply the fix, and retry the job automatically - self-healing is on by default.