Grow/shrink nemesis - growth improvements (For serverlessv2)

[mykaul]

The growth of a cluster should be done in different (random / pre-determined ways)

1. Double the cluster. You have X nodes now (in some topology - multi-DC, etc.) - you'd have the same topology, with 2X nodes. Addition of nodes will be done in parallel [Raft based topology only]
2. Grow in the minimal step - you have X nodes of i4i.8xlarge, you'd add (1 node per AZ ,etc.) i4i.xlarge nodes to it - in parallel. [Raft based topology only + Tablets enabled only]
3. Grow with the same instance type plus 1 node (I believe this is the current functionality).

There should be a way for the user to choose which policy to use for growth (or randomize between them).
We'll need similar support for shrink.

CC @dani-tweig

[fruch]

SCT isn't gonna do any of those calculation on it's own.

user would need to configure the test to it's needs.

we are gonna supply a configuration for the instance type that would be used in grow-shrink nemesis
the amount of nodes in this nemesis is already configurable.

a user would be able to configure both of those base on it's needed and logic he want to test

[fruch]

we wrote it down in QA tasks, few weeks back:

• https://github.com/scylladb/qa-tasks/issues/1738

[mykaul]

@fruch - after implementing https://github.com/scylladb/qa-tasks/issues/1738 - what else remains here?

[fruch]

@fruch - after implementing scylladb/qa-tasks#1738 - what else remains here?

I don't know, i.e. I'm not using this option at all.

@pehala @temichus, if it's enough for the work in serverlessv2 or whatever it's name now, that we can close it.

[pehala]

There are a few PRs open for elastic cloud that make use of that option, I think the subject of this issue is resolved.