devenv_ami and other jenkins jobs should use Docker 1.10 "push as v1" instead of downgrading

[stevekuznetsov]

As seen here:

+ vagrant push-openshift-release --registry docker.io/

exit $exitcode
[ec2-user@ip-172-18-12-220 ~]$ export TERM=vt100
[ec2-user@ip-172-18-12-220 ~]$ stty raw -echo
[ec2-user@ip-172-18-12-220 ~]$ Error while running ssh/sudo command: 
echo "Pushing release images"
set -e
pushd /data/src/github.com/openshift/origin
  OS_PUSH_BASE_IMAGES="" OS_PUSH_BASE_REGISTRY="docker.io/" hack/push-release.sh
popd
The SSH command responded with a non-zero exit status. Vagrant
assumes that this means the command failed. The output for this command
should be in the log above. Please read the output to determine what
went wrong.

@smarterclayton do you understand this?

/cc @jwforres @danmcp

[smarterclayton]

Were there any other logs?

On Wed, Jul 27, 2016 at 2:25 PM, Steve Kuznetsov notifications@github.com
wrote:

As seen here
https://ci.openshift.redhat.com/jenkins/job/devenv_ami/4685/console:

• vagrant push-openshift-release --registry docker.io/

exit $exitcode
[ec2-user@ip-172-18-12-220 ~]$ export TERM=vt100
[ec2-user@ip-172-18-12-220 ~]$ stty raw -echo
[ec2-user@ip-172-18-12-220 ~]$ Error while running ssh/sudo command:
echo "Pushing release images"
set -e
pushd /data/src/github.com/openshift/origin
OS_PUSH_BASE_IMAGES="" OS_PUSH_BASE_REGISTRY="docker.io/" hack/push-release.sh
popd
The SSH command responded with a non-zero exit status. Vagrant
assumes that this means the command failed. The output for this command
should be in the log above. Please read the output to determine what
went wrong.

@smarterclayton https://github.com/smarterclayton do you understand
this?

/cc @jwforres https://github.com/jwforres @danmcp
https://github.com/danmcp

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#10059, or mute the thread
https://github.com/notifications/unsubscribe-auth/ABG_p_ttt9GPpX0RbBCFiO1YBm-138GQks5qZ6KEgaJpZM4JWe6Q
.

[stevekuznetsov]

Not that I can tell. Very weird.

[smarterclayton]

If it's brand new it could be due to my merged changes. Not sure how.

On Wed, Jul 27, 2016 at 2:40 PM, Steve Kuznetsov notifications@github.com
wrote:

Not that I can tell. Very weird.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#10059 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABG_p7aLRukq8QNSP0cSfg0kDDSERQ-qks5qZ6YzgaJpZM4JWe6Q
.

[smarterclayton]

Not positive, but it looks like rebuilding the images on docker 1.9.1 is failing, or not happening

  vagrant build-origin-base-images
  vagrant build-origin --images

Base images works, but build-origin --images does nothing (displays no output). That's deeply suspicious (after downgrade, no images should be present, so builds should work).

Actually... looking at build-origin --images, it's completely wrong.

[smarterclayton]

build-origin --images does not do the right thing, so the short term fix is going to be invoke the correct image build flow on Docker 1.9. The next step would be to stop downgrading to Docker 1.9, and instead use the new flag added to Docker 1.10 to push using v1 schema.

[smarterclayton]

Ok, fixed for now. We now need to switch to stop downgrading 1.9, use 1.10, and use the "push v1".

[smarterclayton]

https://ci.openshift.redhat.com/jenkins/job/devenv_ami/4726/console passed

[smarterclayton]

@bparees FYI, repurposing this.

[bparees]

@smarterclayton are you going to update the push operations done within our various vagrant plugins (eg the s2i image tests/pushes) to do this?

[bparees]

(right now the s2i/db image jenkins jobs explicitly downgrade to 1.9)

[smarterclayton]

I don't know that I am necessarily, but someone needs to, and we should do it for all our jobs. I think we need to update our AMIs first to have the "push as v1" setting. OR we change that setting at the end of the AMI and restart docker.

[bparees]

push as v1 is a dockerconfig setting, or an arg on the docker push command?

[smarterclayton]

/etc/sysconfig/docker flag on startup of the daemon

On Tue, Aug 2, 2016 at 1:13 PM, Ben Parees notifications@github.com wrote:

push as v1 is a dockerconfig setting, or an arg on the docker push command?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#10059 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABG_pwdZiTWLbbq1_OfdP55Tuho6MIkkks5qb3q5gaJpZM4JWe6Q
.

[stevekuznetsov]

That belongs in vagrant-openshift

[bparees]

@stevekuznetsov why don't we want to do it in the ami?

[stevekuznetsov]

I don't understand your question -- we configure /etc/sysconfig/docker in vagrant install-origin-base. We should add this new config option to that preexisting logic.

[stevekuznetsov]

closing to prefer vagrant issue