[chart/redis-ha][BUG] Can only do a rolling update of the ha-proxy deployment on a cluster with min 4 nodes

[martijnvdp]

Describe the bug
Can only do a rolling update of the ha-proxy deployment on a cluster with min 4 nodes
k rollout restart deployment argocd-redis-ha-haproxy
the pod affinity rules and the default spec:

    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%

requires a cluster of 4 nodes to work

changing the deployment spec to :

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1

fixes the issue

To Reproduce
start a rolling update of the haproxy deployment on a cluster with 3 nodes with afinnity rules set
hardAntiAffinity: true

Expected behavior
i expected to be able to do a rolling update on a cluster with less than 4 nodes

Additional context
see also https://stackoverflow.com/questions/65063122/kubernetes-podantiaffinity-affects-deployment-failedscheduling-didnt-match

[martijnvdp]

argocd has this also added to the deployent helm chart
https://github.com/argoproj/argo-helm/blob/68f24d146c3c08e64117d98e3d58a6f073ec27a5/charts/argo-cd/values.yaml#LL141C17-L141C17

[DandyDeveloper]

Hello @martijnvdp

I've been very neglectful of this repo as I've moved overseas back to my home country.

I'll try to jump on your PR ASAP to get it in, I'll hopefully be settled over the coming week or two.

Thanks for your patience.

[v-wan]

Hello @DandyDeveloper, please take a look at the PR. It should be expected to perform a rolling upgrade using the default settings of 3 replicas

[DandyDeveloper]

Fixed in 4.25.1 per #250