Chris Wong

Tests failing are both known broken for commit this is based on

Had this problem, root cause was that compactor was repeatedly failing because it didn't have permission to delete objects in the s3 bucket storage, adding DeleteObject permissions fixed it.

Meta point, why does this file have 6 code owners (and my approval doesn't actually do anything)? I'd be happy to make docs PRs myself as things come up, but...

> Do you want to include any possible actions to fix the situation? I think at the very least it would make sense if I could approve PRs for changes...

^Now that I look at it, I'm not sure if that suggestion was for the code owner thing or for fixing the reconnect issue. For the reconnect issue, there's also...

> RE the reconnect issue, why did the user say " But after re-creating the head pod the its back to normal?" Basically, user needs to re-created the head node...

Seeing a similar problem, looks a lot like #3326, except we're using docker instead of buildah -- after upgrading docker we're seeing blobs that were previously marked as `application/vnd.docker.image.rootfs.diff.tar.gzip` (using...

Mostly grasping at straws, might be related to this string of commits, since seems like the main spots where `application/vnd.oci.image.layer.v1.tar` (i.e. `ocispec.MediaTypeImageLayer`) has been used thats new between 24.0 and...

On my end it looks like this was an issue with bazel docker rules, unrelated to containerd/docker -- https://github.com/bazelbuild/rules_docker/blob/master/container/go/pkg/compat/image.go#L175-L180 It looks like in 25.0 docker started outputting OCI compliant tarballs,...

Ended up having to patch bazel rules docker by adding this after [here](https://github.com/bazelbuild/rules_docker/blob/master/container/go/pkg/compat/image.go#L174): ``` if mediaType == types.OCIUncompressedLayer { // The remote pusher always stores blobs in compressed format, even...