Affects Version/s: 6.23.13
Fix Version/s: None
- Problem description: When we deploy a docker image with same name and tag in parallel, artifactory fails to handle the request due to a race condition and causes 500 Internal server error.
- What is the impact to the customer - What are they trying to achieve? where does that bug hurts them? what does it stop them from achieving? This is impacting the customer's builds as the pushes fail with 500 error, and also the pulls of the same image fail with a 404. Due to this their production builds are getting blocked, as an alternative they have to add retries in each of the docker/podman pushes which is causing all their builds to be slow in production and so having a huge impact.
- What is the expected behavior? Docker pushes should go through without the 500 errors
- *Steps to reproduce:*We can reproduce by having the same docker pull command run simultaneously. I tested with iTerm where I used the Broadcast option to run the same docker pull command in about 5 tabs simultaneously and could see the 500 error in 2 tabs as shown below:
docker push mill.jfrog.info:13256/docker/centos:1.0
The push refers to repository [mill.jfrog.info:13256/docker/centos]
received unexpected HTTP status: 500 Internal Server Error
- Environment details:
- Artifactory version 6.23.13