Make pacemaker commit really ensurable

[Vladimir Kuklin]

We need to rewrite pacemaker libraries to check
* if committed data was really committed by rereading CIB
* if all resources were committed to the cluster configuration
* it still can be committed, if CIB was changed between shadow create and shadow commit.

[Vladimir Kuklin (vkuklin)]

This bug is related to https://bugs.launchpad.net/fuel/+bug/1279211

[Vladimir Kuklin (vkuklin)]

This is also true for cs_shadow resources.

[Sergey Vasilenko (xenolog)]

I think we should implement part of cs_shadow in cs_primitive. If no early-created shadow given.
This guarantees us creating shadow cib directly before creating resource.

[Bogdan Dobrelya (bogdando)]

I agree, it might be a god point, to auto-create shadow CIB, if its name has been referenced by any Corosync entity in corresponding provider (Sergey probably is talking about cs_resource, not cs_primitive).
The similar to some programming languages automatically declare undeclared variables instead of raising an errors...

[Bogdan Dobrelya (bogdando)]

Looks like we could just get rid of cs_commit/cs_shadow instead of trying to make it able to fix merge conflicts for shadow cib on fly (would be a waste of time, I believe), see https://bugs.launchpad.net/fuel/+bug/1283062/comments/7

[Andrew Woodward (xarses)]

Upping to high as there are a couple of duplicates pointed here

[Bogdan Dobrelya (bogdando)]

According to https://review.openstack.org/#/c/101242/ we will remove all cs_shadow and cs_commit from puppet manifests, so this issue should be set to won't fix or invalid

[Dmitry Borodaenko (angdraug)]

I think Won't Fix is a more correct status since we plan to implement a different solution (described by https://blueprints.launchpad.net/fuel/+spec/ha-pacemaker-improvements) that makes this bug as originally defined irrelevant.