Request basic memory for antrea-controller (#6233)

Starting with K8s v1.22, system-cluster-critical Pods no longer get low
OOM score like system-node-critical Pods. antrea-controller's OOM score
adjustment was changed from -997 to 999 passively, which made it just
have a bit higher chance of survival over best-effort Pods (OOM score
adjustment: 1000). While it's hard to say how much memory it can use as
it's highly related to the scale of the cluster, we can add a basic
memory request which can decrease the score a bit according to the
formula for burstable Pods:

  oomScoreAdjust := 1000 - (1000*memoryRequest)/memoryCapacity

It should also be a bit helpful to avoid antrea-controller to be
scheduled on a Node with very few available memory.

Signed-off-by: Quan Tian <[email protected]>

build/charts/antrea/README.md

@@ -73,7 +73,7 @@ Kubernetes: `>= 1.16.0-0`
 | controller.antreaController.extraEnv | object | `{}` | Extra environment variables to be injected into antrea-controller. |
 | controller.antreaController.logFileMaxNum | int | `4` | Max number of log files. |
 | controller.antreaController.logFileMaxSize | int | `100` | Max size in MBs of any single log file. |
-| controller.antreaController.resources | object | `{"requests":{"cpu":"200m"}}` | Resource requests and limits for the antrea-controller container. |
+| controller.antreaController.resources | object | `{"requests":{"cpu":"200m","memory":"100Mi"}}` | Resource requests and limits for the antrea-controller container. |
 | controller.apiNodePort | int | `0` | NodePort for the antrea-controller APIServer to server on. |
 | controller.apiPort | int | `10349` | Port for the antrea-controller APIServer to serve on. |
 | controller.enablePrometheusMetrics | bool | `true` | Enable metrics exposure via Prometheus. |

build/charts/antrea/values.yaml

@@ -373,6 +373,7 @@ controller:
     resources:
       requests:
         cpu: "200m"
+        memory: "100Mi"
 
 flowExporter:
   # -- Enable the flow exporter feature.

build/yamls/antrea-aks.yml

@@ -7044,6 +7044,7 @@ spec:
           resources:
             requests:
               cpu: 200m
+              memory: 100Mi
           command: ["antrea-controller"]
           # Log to both "/var/log/antrea/" and stderr (so "kubectl logs" can work).
           args:

build/yamls/antrea-eks.yml

@@ -7045,6 +7045,7 @@ spec:
           resources:
             requests:
               cpu: 200m
+              memory: 100Mi
           command: ["antrea-controller"]
           # Log to both "/var/log/antrea/" and stderr (so "kubectl logs" can work).
           args:

build/yamls/antrea-gke.yml

@@ -7042,6 +7042,7 @@ spec:
           resources:
             requests:
               cpu: 200m
+              memory: 100Mi
           command: ["antrea-controller"]
           # Log to both "/var/log/antrea/" and stderr (so "kubectl logs" can work).
           args:

build/yamls/antrea-ipsec.yml

@@ -7101,6 +7101,7 @@ spec:
           resources:
             requests:
               cpu: 200m
+              memory: 100Mi
           command: ["antrea-controller"]
           # Log to both "/var/log/antrea/" and stderr (so "kubectl logs" can work).
           args:

build/yamls/antrea.yml

@@ -7042,6 +7042,7 @@ spec:
           resources:
             requests:
               cpu: 200m
+              memory: 100Mi
           command: ["antrea-controller"]
           # Log to both "/var/log/antrea/" and stderr (so "kubectl logs" can work).
           args: