-
Notifications
You must be signed in to change notification settings - Fork 451
Open
Description
Describe the bug
Pulling an image (whether it be the origin, agent or manually) returns the following in the origin logs (/var/log/kraken/kraken-origin/
) after re-deploying kraken. Image fails to pull.
INFO handler/handler.go:98 500 GET /namespace/nvidia/cuda/blobs/sha256:2806d80fe28f9a42dfe4a1b30aeaefb3f75436d3c7fbfc6dfe7e71e1e5d3840a failed to verify data: digests do not match
To Reproduce
Steps to reproduce the behavior:
- Spin up Kraken using the helm chart. Configure GCS as the backend.
- Push an image via the kraken proxy.
- Run a separate deployment that uses this image. Verify it gets pulled and runs correctly.
- Tear down both deployments.
- Re-deploy kraken using helm chart using same configuration as Step 1.
- Run a new deployment using the same yaml file as Step 3.
Check logs of kraken-origin to find the above error.
Expected behavior
Image should pull down to origin(s).
Screenshots
If applicable, add screenshots to help explain your problem.
Environments
- Kubernetes on Google Cloud
- Kraken backend is Google Cloud Storage
- Running Kraken 0.1.4
Additional context
Add any other context about the problem here--environment configurations, logs, etc.
Build index backend configuration:
- namespace: .*
backend:
gcs:
username: kraken-service-account
bucket: gke_image_cache
root_directory: /tags/
name_path: docker_tag
Oirigin backend configuration:
- namespace: .*
backend:
gcs:
username: kraken-service-account
bucket: gke_image_cache
root_directory: /blobs/
name_path: identity
Metadata
Metadata
Assignees
Labels
No labels