Overview over Cluster Settings

[garloff]

As a SCS KaaS developer, I want to have an overview over the settings that various cluster management solutions offer.
https://scs.sovereignit.de/nextcloud/apps/files/?dir=/Sovereign%20Cloud%20Stack/Technology&fileid=5021

• SCS: SCS settings for KaaS #75 - Kurt, Sven, Eduard
• Gridscale: Cluster Settings for Gridscale #76 - Andrej & Ibrahim
• Hetzner: KaaS settings overview for Hetzner #77 - Janis & Sven
• Cloud&Heat: TBD (-> Jonas)
• Tanzu: TBD (-> Toens)
• Gardener: (Shoot) Cluster settings in Gardener #81 - Malte
• GiantSwarm: TBD (Kurt -> Cornelius Keller / )
• Rancher: TBD (? Andrej - possibly outdated / B1)
• Kubermatic: TBD (? B1?)

Definition of Ready:

• User Story is small enough to be finished within one sprint
• User Story is clear and understood by the whole team
• Acceptance criteria are defined
• Acceptance criteria are clear and understood by the whole team

Definition of Done:

• All acceptance criteria are met
• Changes have been reviewed
• CI tests have run successfully
• Documentation has been updated
• Release Notes have been updated

[garloff]

Discussion 2022-05-23:

• Avoid having too high entry-barriers and complexity just to have a base SCS compatibility
• Focus on core!
• Things that can easily be added by the customer afterwards should not be part of the base requirement.
• Example: ingress: Requires load-balancer, which should be required. (Ingress should not be required.)
• Trouble is that some things that can be done afterwards ARE cloud-provider (and/or node image) dependent

[garloff]

Node images discussion:

• Upstream CAPI images exist (but are not meant for production)
• Provider-independent images are very hard to do (hard to avoid provider-specific things)
• So we would to define requirements to images rather than the images themselves
• Still be useful to feed standard requirements upstream

Need discussion with upstream CAPI image folks.

[garloff]

Discussion on node-problem-detector:

• Can be deployed as daemonset, so does not need to be integrated in node image
• daemonset is under user control, so this should be preferred (over the systemd unit approach)
• Requirement for node images would be that they don't break the daemonset way of deploying n-p-d

Trouble is that Node-problem-detector implementation depends on the exact OS (linux distribution) used in the node image.

[fkr]

This issue is marked for R3. Will this be scope of R4?