Bug 1802303

I see that 2 notes as listed in comment 5 has been added to the docs. And that these notes communicates what Eran is suggesting in the description of this bug. 

That said, I don't understand the purpose nor the reasoning behind this request.

What does sentence "to be counted towards subscription" exactly mean?
What is a customer reading "You also need to make sure that the CSI pods are running on the worker nodes." expected to do?
Why would customer have a  CSI pods in Pending state?

I also don't see answer to Laura's question from comment 2.

Do we implement the suggestions as described here in ocs-ci installation?

Sorry but I didn't get what OCS as infra exactly means here as well.
Is this somewhere described defined in our doc as if not this should be defined to let customer understand it correctly.

I tried to find some openshift documentation about infra nodes and usually found labels:
node-role.kubernetes.io/infra=true
For example here:
https://docs.openshift.com/container-platform/3.10/admin_guide/manage_nodes.html#infrastructure-nodes

But you refer to:
node-role.kubernetes.io/infra=''

I think we should be consistent with other doc.

+1 for Martins comment about better explanation of: What does sentence "to be counted towards subscription" exactly mean?

If I am a customer I would like to know differences of running OCS as infra and normal use case where we don't set the above.


About the infra nodes which let me more understand the logic behind that is described for example here:
https://stackoverflow.com/questions/43844511/purpose-of-adding-openshift-intrastructure-nodes

But can we describe which exact pods will be running on those infra pods? Will there be running just some of them or all OCS pods?
Thanks

(In reply to Martin Bukatovic from comment #7)
> Do we implement the suggestions as described here in ocs-ci installation?

In reply to this. We currently do not do anything mentioned above as we usually have 3 masters / 3 worker nodes  and we don't have separated nodes for infra.

Is my understanding correct that for this approach we should have at least 3 masters 3 workers 3 infra nodes?

Is there some minimum requirements for infra nodes from OCS point of view and where is it documented?

The current state of the docs as referenced in comment 5 via preview url doesn't communicate whole context of the problem, and it's unlikely that customers will understand it in it's current form.

>>> ASSIGNED

Note that this is failedqe mainly because of a failed planning (the bug wasn't clarified enough to qualify to get a qe ack).

After evaluation the feedback from Eran and Petr, I believe that this requires:

- more careful thinking and wording to make sure we convey the information as planned (eg. I don't think that the note about particular problem with tainted notes should be listed among installation requirements)
- testing coordination with infra team

And based on that I'm suggesting to drop this out of OCS 4.3.

That said, we could consider this 4.3.z doc only async to get this out asap, when the purpose is more clarified.

Suggesting pushing the content addition asked in this BZ to post 4.3 GA. We need to do better preparation and coordination as Martin mentioned.
Olive, Ken, can we revert the docs changes so they will be included in a docs bactch update / z-stream for 4.3?

I'm not changing the bug status for now.

Dropping qa ack as noted in comment 12, to make sure status of this bug is correct.

Anjana/Rejy, could you please re-target? 
Also, suggest moving to NEW

*** Bug 1823895 has been marked as a duplicate of this bug. ***

All the requested changes are in the doc and looks good. Marking this as verified.

OCS 4.6.0 GA completed on 17 December 2020