enable registry middleware acceptschema2

[aweiteka]

Signed-off-by: Aaron Weitekamp aweiteka@redhat.com

[ncdc]

Also need to change the default here https://github.com/openshift/origin/blob/master/pkg/dockerregistry/server/repositorymiddleware.go#L173

[ncdc]

cc @vbatts

[smarterclayton]

Also need to double check Ansible.

[mfojtik]

@miminar @legionus PTAL

[legionus]

Yep. Also need to change the default in the middleware code and in the ansible

[openshift-bot]

Evaluated for origin testextended up to 304fba4

[openshift-bot]

continuous-integration/openshift-jenkins/testextended FAILURE (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin_extended/65/) (Base Commit: 010f390) (Extended Tests: core(imageapis), core(builds))

[vbatts]

@miminar (carrying forward the conversation of #8938 (comment))

Does "accepting" schema2 mean "forcing" schema2? It does not sound like it would.

[miminar]

Does "accepting" schema2 mean "forcing" schema2? It does not sound like it would.

@vbatts That's correct. It's up to the client, what version of schema it will upload. Setting the acceptschema2 to true only allows for the schema to be accepted - docker (1.10+ will not fall back to schema 1). Older clients or rh docker daemon running with a flag to enforce schema 1 will still push schema 1 regardless of this option.

Do we need acceptschema1 as well? 😄

[miminar]

Apparently, there cannot be two different extended focus statements. Trying one more time with a regexp

[vbatts]

@miminar i would honestly see if we can just jump to allowing v2.s2, skipping v2.s1, and hopefully next will be allowing oci-v1. To address your concern, how worried are we, that a newer client will push v2.s2 content, and then an older client is broken in fetching it?

@mtrmac thoughts on just allowing v2.s2 format and skipping v2.s1 on the openshift registry?

[ncdc]

AFAIK we require docker 1.10, if not 1.12. cc @smarterclayton @sdodson @brenton @eparis to confirm.

[mtrmac]

I would very much like acceptschema2 to be true by default. I don’t see anything about “skipping schema1” in this PR.

[brenton]

@ncdc Here's the docker support matrix for ocp:

3.0, 3.1 = docker 1.8
3.2, 3.3 = docker 1.10
3.4 = docker 1.12

[ncdc]

Thanks @brenton. Given that info, I don't see any need to support schema 1 in our registry any more. Anyone disagree?

[brenton]

I'm pretty sure images from registry.access.redhat.com use v2.s1. Would that no longer supporting this schema cause breakage?

[ncdc]

I don't think we're talking about removing support for v2.s1. We're just talking about allowing v2.s2. Right?

[smarterclayton]

What's the status on his?

[aweiteka]

Based on the comments this is safe to merge.

[aweiteka]

@mfojtik I recommend pulling this into the next release. Sooner than later would be good to have some time with origin testing.

[ncdc]

It'll have to wait until next sprint starts (4/24)

[aweiteka]

@mfojtik ping

[mfojtik]

@legionus @miminar ?

[miminar]

I don't think we're talking about removing support for v2.s1. We're just talking about allowing v2.s2. Right?

That's right. To be precise, this concerns just the manifests created via a push to the integrated registry. We already allow for consuming v2.s2 either by taging or image import.

@aweiteka can you please change false to true here as well?

Hmm, looking at this I believe there's a bit more work todo in our tests. @aweiteka do you perhaps want me to take over?

[testextended][extended:core(builds|imageapis|images)]

[aweiteka]

Hmm, looking at this I believe there's a bit more work todo in our tests. @aweiteka do you perhaps want me to take over?

@miminar that would be great, if you don't mind. :)

[smarterclayton]

This is required for 3.6

[smarterclayton]

[test]

[openshift-bot]

Evaluated for origin test up to 304fba4

[openshift-bot]

continuous-integration/openshift-jenkins/test SUCCESS (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/1924/) (Base Commit: 90592c5)

[smarterclayton]

@dmage do you think you can help get this over the hump since @miminar has a billion things going on?

[dmage]

I'll take a look as soon as I've finished with registry pruning.

re[testextended][extended:core(builds|imageapis|images)]

[openshift-bot]

The Origin testextended job could not be run again for this pull request.

• If the proposed changes in this pull request caused the job to fail, update this pull request with new code to fix the issue(s).
• If flaky tests caused the job to fail, leave a comment with links to the GitHub issue(s) in the openshift/origin repository with the kind/test-flake label that are tracking the flakes. If no issue already exists for the flake you encountered, create one.
• If something else like CI system downtime or maintenance caused the job to fail, contact a member of the Team Project Committers group to trigger the job again.

[legionus]

LGTM

[smarterclayton]

[merge][severity:blocker]

Unless there's a good reason still?

[openshift-bot]

Evaluated for origin merge up to 304fba4

[openshift-bot]

continuous-integration/openshift-jenkins/merge FAILURE (https://ci.openshift.redhat.com/jenkins/job/merge_pull_request_origin/1014/) (Base Commit: ee15d99) (PR Branch Commit: 304fba4) (Extended Tests: blocker)

[smarterclayton]

Daemmonset flake, force merging