The default version of Docker for jobs that use setup_remote_docker has changed from 20.10.23 to 24.0.6.
The vast majority of users should see no impact. If you experience issues, the 20.10.23 version can be specified within your .circleci/config.yml file. Note: 20.10.23 will be deprecated from CircleCI’s platform and moved to an end-of-life state at a soon-to-be-announced date.
The default Docker version is used if a specific version is not specified in your .circleci/config.yml file or if you are using the current or previous tags in the image stanza in your .circleci/config.yml file.
I got this information by Email some days ago, and there was also denoted that I should comment here, if there are any issues…
Well, I would say there are issues, because suddenly from now my docker image build is broken (I didn’t change anything, and it really worked well for many months without any issue…).
I got this error message:
“Step 1/10 : FROM amazoncorretto:17
Get “registry-docker-io v2”: EOF
Exited with code exit status 1”
Respectively this one before:
“Sending build context to Docker daemon 134.8MB
Step 1/10 : FROM amazoncorretto:17-alpine-jdk
Head “DOCKER HUB URL”: Get “AUTH DOCKER URL”: EOF
Exited with code exit status 1”
And to make it worse, I deployed one of my last “successfully” built Docker images (some hours before, this afternoon) to my productive environment, and suddenly the deployment is also broken! (The funny thing is, that there were NO changes in my actual Docker image, it is just also always built due to other changes…).
So I would REALLY appreciate if those issues get fixed…
P.S: This comment function sucks a little bit, due to it don’t let me include the Docker-Hub links from the error message…
Okay and just one further update:
Now even the “normal” builds are broken.
With “normal” builds I mean the builds that don’t produce a Docker image, but builds that use Docker images for building something (like e.g. cimg/openjdk:17.0.3 which is used for producing a JAR file artefact).
So I don’t know if this is related to the Docker upgrade, but something is really broken now…
@Clemens85 were you seeing issues before today? Dockerhub is experiencing degraded service right now which we’ve noticed is impacting jobs using Docker on our platform: https://www.dockerstatus.com/
We’re in the process of upgrading our status page to reflect the degradation of our upstream provider.
I got this error message during the building after updated to docker 24:
Unable to find image ‘tonistiigi/binfmt:latest’ locally
latest: Pulling from tonistiigi/binfmt
docker: Get …/v2/tonistiigi/binfmt/manifests/sha256:6088cbd69c369178ffd6b68992f342c3a9d5c3cc619bbaa4bfe9a98cb23893d0: EOF.
See ‘docker run --help’.
Exited with code exit status 125
Can you guys help me, please?
Oh… This might then be really bad luck. I wasn’t aware of the Dockerhub issues, my first intuition was hence the Docker update on CircleCI, because it was suddenly so heavy broken.
I faced these issues not before today, so it is very likely that it is just the Dockerhub issue. Sorry then for the unnecessary noise.
No worries! We just put up a page on our side to reflect the issues with DockerHub as well.
If the issues persist after the DockerHub issues are resolved, don’t hesitate to comment again or email me at sebastian@circleci.com.
@mateusaugusto did this just start happening today? See below regarding incident with DockerHub that may be causing that issue: CircleCI Status - Issues Pulling Images From DockerHub
Just to come back to you: It was definitely just the Dockerhub issue, and had nothing to do with the CircleCI Docker version upgrade.
So again sorry, should just have looked at first on the Dockerhub status pages.
1 Like