Fix etcd client generation (#10769)

* ci: redefine multinode to node-etcd-client

This should allow to catch several class of problem rather than just
one -> from network plugin such as calico or cilium talking directly to
the etcd.

* Dynamically define etcd host range

This has two benefits:
- We don't play the etcd role twice for no reason
- We have access to the whole cluster (if needed) to use things like
  group_by.
This commit is contained in:
Max Gautier
2024-01-16 15:50:41 +01:00
committed by GitHub
parent a0a2f40295
commit a9e29a9eb2
5 changed files with 28 additions and 25 deletions

View File

@@ -1,6 +1,6 @@
# Node Layouts
There are six node layout types: `default`, `separate`, `ha`, `scale`, `all-in-one`, and `multinode`.
There are six node layout types: `default`, `separate`, `ha`, `scale`, `all-in-one`, and `node-etcd-client`.
`default` is a non-HA two nodes setup with one separate `kube_node`
and the `etcd` group merged with the `kube_control_plane`.
@@ -18,7 +18,8 @@ never actually deployed, but certificates are generated for them.
`all-in-one` layout use a single node for with `kube_control_plane`, `etcd` and `kube_node` merged.
`multinode` layout consists of two separate `kube_node` and a merged single `etcd+kube_control_plane` node.
`node-etcd-client` layout consists of a 4 nodes cluster, all of them in `kube_node`, first 3 in `etcd` and only one `kube_control_plane`.
This is necessary to tests setups requiring that nodes are etcd clients (use of cilium as `network_plugin` for instance)
Note, the canal network plugin deploys flannel as well plus calico policy controller.