klog | Command line tool for time | Command Line Interface library

You're going to want something along the lines:

--logging-format=json is a flag which need to be set on all Kuberentes System Components ( Kubelet, API-Server, Controller-Manager & Scheduler). You can check all flags here.

Unfortunately you cant do it right now with AKS as you have the managed control plane from Microsoft.

The issue is related to SDN not working properly.

As a general rule, troubleshooting this, we would check the SDN pods (calico, weave, or in this case flannel), are they healthy, any errors in their logs, ...

Check iptables (iptables -nL) and ipvs (ipvsadm -l n) configuration nodes.

Restart SDN pods, as well as kube-proxy, if you still didn't find anything.

Now, on this specific case, we're not suffering from an outage: cluster is freshly deployed, it's likely the SDN never worked at all - though this may not be obvious, with a kubeadm deployment, that doesn't ship with other pods than the defaults, most of which using host networking.

The kubeadm init command mentions that pod CIDR is some 192.168.5.0/24, which brings two remarks:

• with all SDN: the pod CIDR is a subnet that will be split into smaller subnets (usually /24 or /25). Each range being statically allocated to Nodes when they first join your cluster

• running flannel SDN: kubeadm init should include a --pod-network-cidr argument that MUST match the subnet configured in the kube-flannel-cfg ConfigMap, see net-conf.json key.

Though I'm unfamiliar with the process of fixing this, there seem to be an answer on ServerFault that gives some instructions, which sounds right: https://serverfault.com/a/977401/293779

Try to change from:

are c libraries built automatically? My assumption is that those are not.

Why? Tests from the offical repo shows, that it is done automatically

is it possible to use cc_library rule to build shared library that can be used in cgo? if so, how to connect those?

See example from here

I don't see anything particularly wrong with the command you are running. However, t2.micro are very small, and may be too small for the cluster to function.

You can have a look at the kops-operator logs why it is not starting. Try kubectl logs kops-controller-xxxx and kubectl describe pod kops-controller-xxx

One of my iop (from kubectl get iop -A) have "policy" or "telemetry" fields which are no longer supported . So I need to delete them.

1. kubectl delete iop [iopname] -n istio-system

This get stucked though because my controller is already gone. So I have to do

1. kubectl edit iop [iopname] -n istio-system
2. Remove istiooperators.install.istio.io from the finalizer
3. Apply

After this I can see no error logs when I tried to install istio-CNI again followed by istioctl operator init.

First of all, thanks to @ShudiptaSharma. His comment helped me in figuring out that I was trying to get the cluster config from outside of the cluster which was leading the program on my local machine (127.0.0.1) from where I am not able to access the cluster.

Further, I tried to figure out how to access the cluster from outside the cluster and found that InClusterConfig is used for running inside cluster use case, when running outside the cluster, something like the following can be used:

I did some calculations and at no point, I see that NLogK better than KlogN.

Since, K <= N, NlogK will always be more than or equal to KlogN.

That does not mean, min heap approach is going to take more time than max heap approch.

Need to consider the below,

1. In Min heap approach, we will update the heap only if the next value is larger than the head. If the array is in ascending order, we will do it (N-K) times, if it is descending order we will not update it at all. On average, the number of times the tree gets updated is considerable less than N.

2. In Max heap, you need to heapify the tree of size N. If K is negligibly small when compared to N, then this time can become a dominant factor. While in the case of min heap, heapify works on the smaller set of K. Also as mentioned in point 1, most of the value of N will not trigger an update of the tree.

I wrote a small program to compare both the approach. The source can be accessed here.

Results for an array ranging from 0 to 1M in random order is below: