OPSEXP-933: Add node selector for deployments on ACS Helm Chart

[Mieszko96]

Ref. OPSEXP-933

[alxgomz]

As I read it. Current implem would make ALL pods scheduled to the same kubernetes nodes. A nodeSelector should be "set-able" for each pod without bringing all the others pods with him.

[Mieszko96]

I don't think if it is necessary for every pod to have their own value. But if you think we want this i can add seperate value for every pod e.g values.imagemagick.nodeSelector

[mteodori]

I don't think if it is necessary for every pod to have their own value. But if you think we want this i can add seperate value for every pod e.g values.imagemagick.nodeSelector

for maximum flexibility I would go as suggested by @alxgomz to have a separate value per each component like the ingress one and a root value for components that are single service in the charts subdir

[alxgomz]

I don't think if it is necessary for every pod to have their own value. But if you think we want this i can add seperate value for every pod e.g values.imagemagick.nodeSelector

You're right that doesn't make sense for every pod. But just because you set a node selector for one, the other shouldn't have it. That's more what I meant.
Nodeselector make sense mainly for:
transformers (to isolate cpu load)
activemq & search & postgres & repo (for pv in simplistic config)
other thatn that... that would be just for more flexibility

[Mieszko96]

Tested every deployment except deployment-ai-transformer.yaml
With extra values.

activemq:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
imagemagick:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
libreoffice:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
pdfrenderer:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
repository:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
share:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
tika:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
transformmisc:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
alfresco-digital-workspace:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
filestore:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
transformrouter:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
alfresco-search:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
alfresco-insight-zeppelin:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
insightzeppelin:
enabled: True
alfresco-sync-service:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
postgresql:
primary:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}
postgresql-syncservice:
primary:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}

ooi:
enabled: True
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}

msTeams:
enabled: True
msTeamsService:
nodeSelector: {"kubernetes.io/hostname": multinode-demo-m02}

[scila1996]

Please add their parameter

NodeAffinity
PodAffinity
Pod Tolerence

[gionn]

@scila1996 can you please open a new issue, describing why do you need them (which is your use case and why nodeSelector is insufficient)? Thanks.

[alxgomz]

LGTM