-
Notifications
You must be signed in to change notification settings - Fork 0
Description
The component currently provides a number of default Helm values for installing Cilium, cf.
component-cilium/class/defaults.yml
Lines 13 to 39 in e38b95c
| cilium_helm_values: | |
| cni: | |
| binPath: /var/lib/cni/bin | |
| confPath: /var/run/multus/cni/net.d | |
| endpointRoutes: | |
| enabled: true | |
| hubble: | |
| tls: | |
| enabled: false | |
| ipam: | |
| mode: cluster-pool | |
| operator: | |
| clusterPoolIPv4MaskSize: "23" | |
| clusterPoolIPv4PodCIDR: 10.128.0.0/14 | |
| kubeProxyReplacement: probe | |
| nativeRoutingCIDR: 10.128.0.0/14 | |
| prometheus: | |
| serviceMonitor: | |
| enabled: false | |
| operator: | |
| resources: | |
| requests: | |
| cpu: 100m | |
| memory: 250Mi | |
| limits: | |
| cpu: 100m | |
| memory: 250Mi |
Some of these values are only valid for OpenShift 4, for example cni.confPath: /var/run/multus/cni/net.d. Other values are inspired by the OpenShift 4 defaults, such as the default Pod CIDR. However, the Pod CIDR values don't seem to match the OCP4.9 defaults completely (OCP4.9 uses 10.128.0.0/10 by default, not the component's 10.128.0.0/14). In contrast, the Cilium Helm chart defaults to 10.0.0.0/8 for the Pod CIDR.
Actual Behavior
I would expect that the component only provides generally applicable default configs. Currently the component must be reconfigured to be usable on non-OCP4 (or technically, non-multus setups).
Expected Behavior
I'd expect that the component defaults don't contain OpenShift 4-specific configuration and that the component documents a minimal configuration required for OpenShift 4.