Kubernetes kubectl shows pods restarts as zero but pods age has changed - kubernetes

Question

Can somebody explain why the following command shows that there have been no restarts but the age is 2 hours when it was started 17 days ago
kubectl get pod -o wide
NAME READY STATUS RESTARTS AGE IP NODE
api-depl-nm-xxx 1/1 Running 0 17d xxx.xxx.xxx.xxx ip-xxx-xxx-xxx-xxx.eu-west-1.compute.internal
ei-depl-nm-xxx 1/1 Running 0 2h xxx.xxx.xxx.xxx ip-xxx-xxx-xxx-xxx.eu-west-1.compute.internal
jenkins-depl-nm-xxx 1/1 Running 0 2h xxx.xxx.xxx.xxx ip-xxx-xxx-xxx-xxx.eu-west-1.compute.internal
The deployments have been running for 17 days:
kubectl get deploy -o wide
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE CONTAINER(S) IMAGE(S) SELECTOR
api-depl-nm 1 1 1 1 17d api-depl-nm xxx name=api-depl-nm
ei-depl-nm 1 1 1 1 17d ei-depl-nm xxx name=ei-depl-nm
jenkins-depl-nm 1 1 1 1 17d jenkins-depl-nm xxx name=jenkins-depl-nm
The start time was 2 hours ago:
kubectl describe po ei-depl-nm-xxx | grep Start
Start Time: Tue, 24 Jul 2018 09:07:05 +0100
Started: Tue, 24 Jul 2018 09:10:33 +0100
The application logs show it restarted.
So why is the restarts 0?
Updated with more information as a response to answer.
I may be wrong but I don't think the deployment was updated or scaled it certainly was not done be me and no one else has access to the system.
kubectl describe deployment ei-depl-nm
...
CreationTimestamp: Fri, 06 Jul 2018 17:06:24 +0100
Labels: name=ei-depl-nm
...
Replicas: 1 desired | 1 updated | 1 total | 1 available | 0 unavailable
StrategyType: RollingUpdate
...
Conditions:
Type Status Reason
---- ------ ------
Available True MinimumReplicasAvailable
OldReplicaSets: <none>
NewReplicaSet: ei-depl-nm-xxx (1/1 replicas created)
Events: <none>
I may be wrong but I don't think the worker node was restarted or shut down
kubectl describe nodes ip-xxx.eu-west-1.compute.internal
Taints: <none>
CreationTimestamp: Fri, 06 Jul 2018 16:39:40 +0100
Conditions:
Type Status LastHeartbeatTime LastTransitionTime Reason Message
---- ------ ----------------- ------------------ ------ -------
NetworkUnavailable False Fri, 06 Jul 2018 16:39:45 +0100 Fri, 06 Jul 2018 16:39:45 +0100 RouteCreated RouteController created a route
OutOfDisk False Wed, 25 Jul 2018 16:30:36 +0100 Fri, 06 Jul 2018 16:39:40 +0100 KubeletHasSufficientDisk kubelet has sufficient disk space available
MemoryPressure False Wed, 25 Jul 2018 16:30:36 +0100 Wed, 25 Jul 2018 02:23:01 +0100 KubeletHasSufficientMemory kubelet has sufficient memory available
DiskPressure False Wed, 25 Jul 2018 16:30:36 +0100 Wed, 25 Jul 2018 02:23:01 +0100 KubeletHasNoDiskPressure kubelet has no disk pressure
Ready True Wed, 25 Jul 2018 16:30:36 +0100 Wed, 25 Jul 2018 02:23:11 +0100 KubeletReady kubelet is posting ready status
......
Non-terminated Pods: (4 in total)
Namespace Name CPU Requests CPU Limits Memory Requests Memory Limits
--------- ---- ------------ ---------- --------------- -------------
default ei-depl-nm-xxx 100m (5%) 0 (0%) 0 (0%) 0 (0%)
default jenkins-depl-nm-xxx 100m (5%) 0 (0%) 0 (0%) 0 (0%)
kube-system kube-dns-xxx 260m (13%) 0 (0%) 110Mi (1%) 170Mi (2%)
kube-system kube-proxy-ip-xxx.eu-west-1.compute.internal 100m (5%) 0 (0%) 0 (0%) 0 (0%)
Allocated resources:
(Total limits may be over 100 percent, i.e., overcommitted.)
CPU Requests CPU Limits Memory Requests Memory Limits
------------ ---------- --------------- -------------
560m (28%) 0 (0%) 110Mi (1%) 170Mi (2%)
Events: <none>

Answer 1

There are two things that might happen:
The deployment was updated or scaled:
age of deployment does not change
new ReplicaSet is created, old ReplicaSet is deleted. You can check it by running
$ kubectl describe deployment <deployment_name>
...
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 1m deployment-controller Scaled up replica set testdep1-75488876f6 to 1
Normal ScalingReplicaSet 1m deployment-controller Scaled down replica set testdep1-d4884df5f to 0
pods created by old ReplicaSet are terminated, new ReplicaSet created brand new pod with restarts 0 and age 0 sec.
Worker node was restarted or shut down.
Pod on old worker node disappears
Scheduler creates a brand new pod on the first available node (it can be the same node after reboot) with restarts 0 and age 0 sec.
You can check the node start events by running
kubectl describe nodes <node_name>
...
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Starting 32s kubelet, <node-name> Starting kubelet.
Normal NodeHasSufficientPID 31s (x5 over 32s) kubelet, <node-name> Node <node-name> status is now: NodeHasSufficientPID
Normal NodeAllocatableEnforced 31s kubelet, <node-name> Updated Node Allocatable limit across pods
Normal NodeHasSufficientDisk 30s (x6 over 32s) kubelet, <node-name> Node <node-name> status is now: NodeHasSufficientDisk
Normal NodeHasSufficientMemory 30s (x6 over 32s) kubelet, <node-name> Node <node-name> status is now: NodeHasSufficientMemory
Normal NodeHasNoDiskPressure 30s (x6 over 32s) kubelet, <node-name> Node <node-name> status is now: NodeHasNoDiskPressure
Normal Starting 10s kube-proxy, <node-name> Starting kube-proxy.