delete successfully pushed message

[QiWang19]

close #1742
Delete the line: Successfully pushed...

Signed-off-by: Qi Wang [email protected]

[vrothberg]

Now it's reporting docker://localhost:5000/alpine:latest@sha256:294c4cf3a9a174563142b85737afca6e14d2ef5eaaed9b7cac13cac787c06e25, which is much better but we currently do not support the image:tag@digest notation. So if a user would copy the output and tries to use it, they will face a parsing error. That might not be the best experience.

I see two options (maybe there are more that I don't yet see):

1. We support parsing image:tag@digest. @mtrmac should weigh in here to be sure I am not missing something.
2. We remove the entire Printf(...) line (as Podman).

[QiWang19]

I remove the Printf(...) for now to keep it same as Podman

[vrothberg]

LGTM, thanks @QiWang19 👍

@TomSweeneyRedHat @nalind @rhatdan PTAL

[TomSweeneyRedHat]

Change is fine, but the commit title and description need to be touched up to reflect the final changes.

[mtrmac]

1. We support parsing image:tag@digest. @mtrmac should weigh in here to be sure I am not missing something.

See earlier discussion in #1407 and linked issues.

[QiWang19]

@vrothberg @TomSweeneyRedHat PTAL

[vrothberg]

LGTM
Thanks for taking care of it!

[TomSweeneyRedHat]

LGTM assuming happy tests

[TomSweeneyRedHat]

@rh-atomic-bot r+

[rh-atomic-bot]

📌 Commit ee6be6b has been approved by TomSweeneyRedHat

[rh-atomic-bot]

⌛ Testing commit ee6be6b with merge a74bdd3...

[rh-atomic-bot]

☀️ Test successful - status-papr, status-travis
Approved by: TomSweeneyRedHat
Pushing a74bdd3 to master...

[wwilson]

Following this change, what's the recommended way to atomically determine the remote digest of a newly-pushed image? We had come to rely on this log message for content addressability, in order to work around the issue that RepoDigest and local image digests do not in general match (containers/skopeo#469).

An alternate workaround would be to skopeo inspect the remote image after push completes, but that's obviously asking for concurrency problems if somebody else uploads an image with the same tag at the same time.

Is there another way that Buildah reports the RemoteDigest of the image it has just pushed? Or should we be going about this a completely different way?

[rhatdan]

@vrothberg @mtrmac @nalind PTAL

[TomSweeneyRedHat]

If you run 'buildah --debug push ...` you get some of that information in the output:

Writing manifest to image destination
DEBU[0001] PUT https://registry-1.docker.io/v2/tomsweeneyredhat/testing/manifests/first 
Storing signatures
DEBU[0002] pushed image "docker.io/tomsweeneyredhat/testing:first@sha256:b552b401d9977f4da64e410727e84a736be5b78f96b561a9b21579b3d1d74a8f" with digest sha256:b552b401d9977f4da64e410727e84a736be5b78f96b561a9b21579b3d1d74a8f 

I'm not sure if that's sufficient for your needs nor do I know of another way to easily do it other than what you've mentioned.

I've just "re-added" the output as a debug line in #1799. If the other folks don't have a better thought, we could go with that. The output using the --debug as shown above would look like:

Writing manifest to image destination
DEBU[0001] PUT https://registry-1.docker.io/v2/tomsweeneyredhat/testing/manifests/first 
Storing signatures
DEBU[0002] pushed image "docker.io/tomsweeneyredhat/testing:first@sha256:b552b401d9977f4da64e410727e84a736be5b78f96b561a9b21579b3d1d74a8f" with digest sha256:b552b401d9977f4da64e410727e84a736be5b78f96b561a9b21579b3d1d74a8f 
DEBU[0002] Successfully pushed //tomsweeneyredhat/testing:first@sha256:b552b401d9977f4da64e410727e84a736be5b78f96b561a9b21579b3d1d74a8f 

There's a lot more output this way though, would this work for you?

[wwilson]

Thanks, I didn't know about the buildah --debug push output. I think (after your change) we could skate by with taking that and feeding it into sed/awk, but I'm a little nervous parsing debug output like that since it seems like the kind of thing that can change without warning (as indeed it did in this case).

If you agree that this is a reasonable thing for people to want to know after pushing an image, should we also think about a more permanent and documented solution?

[TomSweeneyRedHat]

I'd like to hear from @nalind and @rhatdan on their thoughts as they had wanted to remove this line. Another approach that we might consider is adding a --verbose option which would print that line as it had previously done if --verbose or -v was passed in ala buildah push -v ...

FWIW, the --debug option can be placed after buildah and before any buildah command to get a lot of insight into what's going on under the covers with a particular command.

[mtrmac]

IIRC removing this text was not an explicit goal, it was just easier than to agree on a format that makes sense when we didn’t know of an user and didn’t have a strong need. (Let’s worry about that part in #1799.)

Originally the functionality was added via containers/image#518 for OpenShift use (I guess via the API) and to the CLI output via #1097 , which does not explicitly say why it was added to the CLI).

---

I mean, if the digest is supposed to be reliably consumable by scripts, we don’t want the scripts scraping the CLI output if we can help it; the CLI could have a buildah push --digestfile (similar to buildah commit --iidfile), or there could be some other structured output format.

--verbose/--debug are easier to implement but ultimately less reliable.

[rhatdan]

Ok, if users want this information, then --digestfile would work. @wwilson WDYT?

[wwilson]

I agree that structured output would be best. —digestfile sounds great.