You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our AKS clusters publick8s and privatek8s are not directly pointed, but since they both used a combination of LB outbound and NAT gateway due to legacy reasons, it will have an effect and we risk going back to SNAT exhaustion issues.
Additionally, these clusters have "legacy" setup which can be improved:
Control Plane API accesses: their APIs are public with IP restrictions which makes it tedious to maintain for admin accesses. We have/had others clusters (infraci-agents-1 and cijio-agents-1) with the "private" access for API which provides automatic DNS: way easier to maintain and restrict through VPN, with less risk of mistaken exposure (not public).
Node pools are using Ubuntu while Microsoft provides Azure Linux which is a lightweight system aimed at running containers in AKS (faster to start, automatic patch upgrades, etc.)
We should re-create these clusters with the new set of settings in order to improve the maintenance and safety of these 2 clusters