Bug 1348324

*** Bug 1348323 has been marked as a duplicate of this bug. ***

Customer used rpm method for installation.

I spoke with Devan (Eng) offline; the procedure should be valid for 3.2 to 3.1 downgrades for RPM-based installations. We did not write a containerized downgrade procedure previously because 3.0 did not support containerized, so that would have to be new docs for containerized downgrade of 3.2 -> 3.1. In the meantime, we can update for just the RPM-based procedure.

Worked with Devan to update the steps, and I tested it out w/ the embedded etcd path; it seemed to work fine. I haven't tested the external etcd path.

PR is here:

https://github.com/openshift/openshift-docs/pull/2351

Preview build is here (internal):

http://file.rdu.redhat.com/~adellape/062416/downgrade32/install_config/downgrade.html

@anli - Can you QE the above preview build as you did originally for 3.1 to 3.0 in https://trello.com/c/6VqWWpm8, this time for 3.2 to 3.1? And if you're able, would you be able to provide an estimate on how long it might take for you to get to?

Thanks!

Just noticed the above PR/build doesn't include downgrading back to Docker 1.8.2 from 1.9.1.

@alex, Yes, we need docker downgrade part. Depend on customer, it should be 1.10 to docker 1.8 or docker 1.9 to docker 1.8. 
For other part, the procedure has no changes.

Can we get an ETA in terms of testing the procedure and update to the documented steps? The customer is looking to get the new version in production to deal with instability.

Anping, can you QE the Docker downgrade steps that I wrote in the following KBase article?

https://access.redhat.com/solutions/2214221

Look for the following heading specifically, under the Resolution section:

"If I updated to Docker 1.9+, how can I downgrade to 1.8.2?"

If you verify those steps, I can add those to my 3.2 -> 3.1 downgrade topic/PR, substituting in the following for the yum swap command:

# yum swap docker-* docker-*1.9.1

I feel like the BZ here would then be satisfied, yes?

(In reply to rfoyle from comment #8)
> Can we get an ETA in terms of testing the procedure and update to the
> documented steps? The customer is looking to get the new version in
> production to deal with instability.

I am able to get back to this in earnest tomorrow. If the BZ passes QE, I could have the docs updated by EOD Friday as well. Otherwise, it would more likely be next Tuesday (due to the upcoming holiday weekend).

The docker downgrade works https://access.redhat.com/solutions/2214221 .

May be we still need to remove one option draw in y OSE 3.2.1
sed -i 's/ --log-driver=json-file --log-opt max-size=50m//' /etc/sysconfig/docker

(In reply to Anping Li from comment #11)
> The docker downgrade works https://access.redhat.com/solutions/2214221 .
> 
> May be we still need to remove one option draw in y OSE 3.2.1
> sed -i 's/ --log-driver=json-file --log-opt max-size=50m//'
> /etc/sysconfig/docker

Those options work with docker-1.8.2 but I've asked the docker team if they're safe to use there or not. For now lets leave them as they are.

@scott, agree.  I will move status to verified.

Anping, I've now added the following to the doc:

- Docker downgrade steps
- Steps on backing up and restoring node-config.yaml
- A warning that this procedure assumes entire cluster downtime

Because the procedure involves entire cluster downtime (and includes a warning), the Docker downgrade steps no longer involve evacuating nodes.

So here is a new preview build:

http://file.rdu.redhat.com/~adellape/062416/downgrade32/install_config/downgrade.html

Can you verify the doc again? Then if it's all good, I'll publish. Thank you!

Shall we add instructions for API Service and Controller service? if it was installed by openshift-ansible, the instructions are as following.

1) Shutting Down the Cluster
 For Native HA masters:
   systemctl stop atomic-openshift-master-api
   systemctl stop atomic-openshift-master-controllers

2) Bringing OpenShift Services Back Online

 For Native HA masters:
   systemctl enable atomic-openshift-master-api
   systemctl enable atomic-openshift-master-controllers
   systemctl start atomic-openshift-master-api
   systemctl start atomic-openshift-master-controllers

BTW:
/usr/lib/systemd/system/atomic-openshift-master-api.service 
/etc/sysconfig/atomic-openshift-master-api
/usr/lib/systemd/system/atomic-openshift-master-controller.service 
/etc/sysconfig/atomic-openshift-master-controller

The above configure file won't be removed by 'yum erase atomic-openshift-master', and only loglevel was changed during ugprade from v3.1 to v3.2. so we can re-use them without any change.

#Adding Additional etcd Members
"member update 511b7fb6cc0001 https://172.18.1.18:2379" should be "member update 511b7fb6cc0001 https://172.18.1.18:2380"

Thanks! PR and preview updated per suggestions/fixes:

http://file.rdu.redhat.com/~adellape/062416/downgrade32/install_config/downgrade.html

Looks good to me.

Commit pushed to master at https://github.com/openshift/openshift-docs

https://github.com/openshift/openshift-docs/commit/4a7d41d74c09a2abd5e6198b480d2e03419318cf
Merge pull request #2351 from adellape/downgrade32

Bug 1348324: Update Downgrading topic for OSE 3.2 -> 3.1

Published on docs.openshift.com:

https://docs.openshift.com/enterprise/3.2/welcome/revhistory_full.html#fri-jul-08-2016

https://docs.openshift.com/enterprise/3.2/install_config/downgrade.html#install-config-downgrade

And on the Customer Portal:

https://access.redhat.com/documentation/en/openshift-enterprise/version-3.2/installation-and-configuration/#fri-jul-08-2016

https://access.redhat.com/documentation/en/openshift-enterprise/3.2/installation-and-configuration/chapter-4-downgrading-openshift