cgroup containerd k8s

tut/k8s/wasted/README.md

@@ -7,12 +7,13 @@ k apply -f wasted.yml
 
 ```
 
-it should be killed and an attempt to restart will occur, then CrashLoopBackOff
+it should be killed by container runtime (containerd), then k8s comes and replaced the status with Crashbackloop (the OOMKilled status will be replaced with Crashbackloop ).
+If no other monitoring were there to record the OOM moment and nobody sees when it happens, like below(last until K8S perform the check and replaced status = Crashbackloop):
 ```bash
 NAME                                       READY   STATUS      RESTARTS      AGE
 memory-usage-deployment-7d787c4d74-krs4n   0/1     OOMKilled   1 (10s ago)   39s
 ```
-
+it may not be obvious why the pod is in Crashbackloop state since there is no k8s event to record the observation of OOMKilled(refer to k8s code below).
 
 clean up
 ```bash
@@ -22,3 +23,5 @@ k delete deployment memory-usage-deployment -n musage
 
 
 https://github.com/kubernetes/kubernetes/commit/4e20a8f52bcce054459f4df537c12e889a02b86c
+
+https://github.com/kubernetes/kubernetes/blob/3ec6a387955b1240ad6d795663513f1ee12ceaec/pkg/kubelet/kuberuntime/helpers.go#L159

vs/java/cgroup/README.md

@@ -227,3 +227,5 @@ https://github.com/KimMachineGun/automemlimit
 https://github.com/google/cadvisor
 
 https://github.com/kubernetes/kubernetes/commit/4e20a8f52bcce054459f4df537c12e889a02b86c
+
+https://github.com/kubernetes/kubernetes/blob/3ec6a387955b1240ad6d795663513f1ee12ceaec/pkg/kubelet/kuberuntime/helpers.go#L159