Docker "No such image" (after prune mid-build) in CI

By Daniel Zoghalchali·Latchkey

A docker run, docker tag, or docker save that references an image by id or tag fails with "No such image" when that image was deleted out from under it - most often by a concurrent docker system prune or docker image prune in another step on a shared daemon.

What this error means

A step fails with Error response from daemon: No such image: myorg/app:ci even though an earlier step built or pulled it. A prune ran in between.

Error response from daemon: No such image: myorg/app:ci

Common causes

A concurrent prune on a shared daemon

A cleanup job running docker system prune -a removed the image while another job still referenced it.

A prune step ordered before the consumer

A workflow that prunes before the step that needs the image deletes it prematurely.

How to fix it

Stop pruning shared images mid-pipeline

1. Move prune to the very end of the pipeline, after all consumers.
2. Avoid -a (remove-all) prune while other jobs share the daemon.

# build/test first, prune last
docker build -t myorg/app:ci .
docker run --rm myorg/app:ci pytest
docker image prune -f --filter "until=1h"

Isolate the daemon per job

1. Use a dedicated dind daemon per job so prunes never affect other jobs.

services:
  docker:
    image: docker:27-dind

How to prevent it

• Order prune steps after all image consumers; scope with --filter until=.
• Give each job an isolated daemon when sharing causes cross-job deletes.

Frequently asked questions

Related guides

References

• Docker documentation
• Docker build cache
• GitHub Actions documentation