use stable/cloud repo for cloud tenants

[stevenGravy]

Currently the node install uses the major version, not checking to see if it should use the cloud/stable.

[r0mant]

@stevenGravy We should not be implementing this logic in the script. Whether to use stable/cloud repo is controlled by the web handlers, e.g.

https://github.com/gravitational/teleport/blob/v13.0.2/lib/web/join_tokens.go#L212

Looks like we may just need to remove h.ClusterFeatures.AutomaticUpgrades condition to always use this channel for all Cloud customers. This change needs to be applied to all other script handlers as well though:

https://github.com/gravitational/teleport/blob/v13.0.2/lib/web/join_tokens.go#L256
https://github.com/gravitational/teleport/blob/v13.0.2/lib/web/join_tokens.go#L286
https://github.com/gravitational/teleport/blob/v13.0.2/lib/web/apiserver.go#L1594

@marcoandredinis Is that right? Are there any other places that need changing?

@marcoandredinis @hugoShaka There shouldn't be any downside in using this channel now for all Cloud customers, is there? We won't actually install the "updater" package unless the Cloud tenant is actually enrolled in auto-upgrades, correct?

[marcoandredinis]

Is that right? Are there any other places that need changing?

Yes, those are the locations we need to change

There shouldn't be any downside in using this channel now for all Cloud customers, is there? We won't actually install the "updater" package unless the Cloud tenant is actually enrolled in auto-upgrades, correct?

Even without automatic upgrades, do we want users to use stable/cloud repo for all their installations?
If so, when operators manually update their agents (node, app and database agents), they will always get the latest release from that channel.
That version should be always in sync with their current auth/proxy.
My only concern is the timeline between updating the stable/cloud channel and the 3 upgrade windows we have for cloud.
If we update the stable/cloud channel before we start the migration, bad things can happen.
If we update the stable/cloud after the latest window, then I think we are good 👍

[r0mant]

@marcoandredinis We only update stable/cloud after Cloud control plane has been upgraded (using this workflow @fheinecke built) so we should be good.

@stevenGravy We'll take over this PR. @marcoandredinis Can you please make necessary changes?

[marcoandredinis]

@hugoShaka @stevenGravy Can you please take a look?

[stevenGravy]

@hugoShaka @stevenGravy Can you please take a look?

Looks right to me!

[public-teleport-github-review-bot]

@stevenGravy See the table below for backport results.

Branch	Result
branch/v12	Failed
branch/v13	Create PR