Clarifications on 'no downtime' during upgrades

[renshuki]

Current wording suggests that no downtime is expected during upgrades whatever the deployment topology is.
Non-HA deployments (on single AZ) are subject to downtime as the Elasticsearch node upgrade happens inline with a rolling change.

Also for Kibana upgrade, all Kibana instances are shutdown simultaneously which makes Kibana inaccessible during the Kibana upgrade plan whatever the zone distribution is.

Preview here:

• Availability during upgrades

[eedugon]

@shainaraskas , this looks good to me, just a few comments for your review:

• We could link to the doc that describes HA and non-HA clusters in ECH.
• We could link to Kibana upgrade description in the new comment about Kibana, where it's also explained that Kibana rolling-upgrade is not an option (Kibana upgrade implies Kibana downtime).

[shainaraskas]

to address @eedugon's comments:

• linking, or being more precise about what the minimum node configuration is, is a good idea
• kibana link: we could be more precise about the kibana downtime thing, but the other upgrade doc is specific to self-managed might be misleading. Edu, in future, we could maybe use a snippet to reuse this info if you think it's helpful for the other deployment types

provided some suggestions to address this stuff

[eedugon]

As agreed with @renshuki , we'll do some extra refinement and share with you the results before merging.
Thanks!

[eedugon]

@shainaraskas : I'm re-requesting your review here, I have adapted your changes and included all the information related with this PR (very important in my opinion (thanks @renshuki ) ) in a dedicated section of the doc.

I have also updated the link to HA in ECH to the document we recently added in the other PR :) As we are planning to improve that one soon, better to link to that from this ECH upgrade doc.

@renshuki : let us know how you feel it, and if it covers your original goal also.

Preview here:

• Availability during upgrades

[shainaraskas]

I really like this approach! thank you both

cc: @eedugon just in case gh won't notify you automatically

[renshuki]

@renshuki : let us know how you feel it, and if it covers your original goal also.

Thank you @shainaraskas and @eedugon for the great rework on this PR, I have nothing to add on my side 😄 All the points I originally mentioned are covered.