GitLab

Resolve "On alpine 3.16 k3s miss cni plugin "flannel""

Closes #33

See merge request !28

Happens on alpine 3.16

Resolve "Give possibility to instantiate a "naked" cluster"

Closes #31

See merge request !26

The objective is to have a functional cluster with almost no pod running on.
User will be able to easily install things like "laboite"

A user might want to deploy a cluster without default ingress and install
its own ingress later.

This solves the following problem:
The reverse proxy waits for all nodes to be up
The nodes wait for the gateway (revProx) to do some DNS to access OpenNebula
(We are in the case where the gateway is the only machine to know how to
contact the openNebula's API)

Resolve "Add routing and DNS server features to k3s reverseproxy role"

Closes #29

See merge request !23

Otherwise, https traffic is blocked

Resolve "Pass the names of the nodes through contextualization"

Closes #26

See merge request !21

This will fix the problem where hostname != VM.NAME
Also, fix a potential infinite loop
Also, fix a typo

Resolve "Use hub.eole.education as proxyhub"

Closes #25

See merge request !20

Resolve "Traefik does not work in cluster after deploy"

Closes #16, #20, #21, and #22

See merge request !17

Use at least 3 replias of traefik pod.

using $ ins not a good idea

fix(cert-manager): Updating cert-manager for traefik

We need to update and improve cert-manager support for traefik

fix(haproxy): updating haproxy configuration with log support

feat(postdeploy): enabling traefik dashboard

fix(postdeploy): installing certmanager ...for real this time

feat(ingress): adding nginx ingress support

fix(openebs): adding missing cstor-csi-disk

fix(bootstrap): improve cluster init to be more accurate.

fix(scipts): shell format cleanup

Resolve "OpenEBS : configuration missing."

Closes #23

See merge request !19

Resolve "HAProxy conf is incomplete and invalid"

Closes #17

See merge request !16

Merge branch '18-with-a-reverse-proxy-the-reverse-proxy-server-should-not-use-haproxy-and-k3s-at-the-same-time' into 'develop'

Resolve "With a reverse proxy, the reverse proxy server should not use haproxy and k3s at the same time"

Closes #18

See merge request !15

Create haproxy config file after all server nodes are instanciated
Create valid names for backend servers
Fix "stats" endpoint
Check port 80 of servers

Fixes #17

Always install haproxy since it isn't a flavor anymore.
Setup/launch either k3s or haproxy, not both of them.

feat(service): adding reverse proxy role setup

Closes #14

See merge request !14

Fixes #14

fix(service): changin the way to detect agents

Closes #11 and #13

See merge request !13

Service has now a leader to support multiple architectures including
multi-server ones.

If the leader has 2 network interfaces we use the second one for the
cluster, otherwise we use the first one. This way we can support both
planned architectures.

Fixes #13

the agents name don't start with agent every time but they allways
contains the word 'agent'

Resolves #11

docs(architecture): adding architecture documentation

See merge request !12

feat(service): adding opennebula service definition

See merge request !11