google_containers

Different levels of availability can be achieved in Kubernetes, it all depends on your requirements.

Your use case seem to be that only one replica should be active at the time to the database.

Even if you use a single replica in a Kubernetes Deployment or StatefulSet, it is regularly probed, using your declared LivenessProbe and ReadinessProbe.

If your app does not respond on LivenessProbe, a new pod will be created immediately.

Since only one replica at a time should have an active connection to your database, a leader election solution is viable.

The passive replicas, that currently does not have the lock, should regularly try to get the lock - so that they get active in case the old active pod has died. How this is done depend on the implementation and configuration.

If you want that only the active Pod in a multiplie replica solution should query the database, the app must first check if it has the lock (e.g. is the active instance).

There is not much difference between a single replica Deployment and a multi replica Deployment using leader election. There might be small differences in the time a failover takes.

For a single replica solution, you might consider using a StatefulSet instead of a Deployment due to different behavior when a node becomes unreachable.

As mentioned in the logs

As posted by user @David Maze:

What's the exact URL you're trying to connect to? What error do you get? (On the load-balancer-autoscaler service, the targetPort needs to match the name or number of a ports: in the pod, or you could just change the hpa-example service to type: LoadBalancer.)

I reproduced your scenario and found out issue in your configuration that could deny your ability to connect to this Deployment.

From the perspective of Deployment and Service of type NodePort everything seems to work okay.

If it comes to the Service of type LoadBalancer on the other hand:

Assuming the minor syntactic errors have been fixed in the shown sample, the following produces the result you're expecting:

This is a very common issue and it required from me a full migration of CIDR IPs.

Most probably, this issue about the overlap of CIDRs between Pods CIDR ( which is IP pool used to assign IPs for services and pods) and CIDR of your network.

in this case, route tables of each node (VM) will ensure that:

This context deadline exceeded generally happens because of

1. Using wrong certificates. You could be using peer certificates instead of client certificates. You need to check the Kubernetes API Server parameters which will tell you where are the client certificates located because Kubernetes API Server is a client to ETCD. Then you can use those same certificates in the etcdctl command from the node.

2. The etcd cluster is not operational anymore because peer members are down.

If you have production or 'live' workloads, the best safe approach is to provision a new cluster and switch the workloads gradually.

Kubernetes keeps its state in etcd so you could potentially connect to etcd and clear the 'drained' state but you will probably have to look at the source code and see where that happens and where the specific key/values are stored in etcd.

The logs that you shared are basically showing that the kube-apiserver cannot start so it's likely that it's trying to connect to etcd/startup and etcd is telling it: "you cannot start on this node because it has been drained".

The typical startup sequence for the masters is something like this:

• etcd
• kube-apiserver
• kube-controller-manager
• kube-scheduler

You can also follow any guide to connect to etcd and see if you can troubleshoot any further. For example, this one. Then you could examine/delete some of the node keys at your own risk:

Please check the Kubernetes version you launched with Minikube.

Spark v2.4.5 fabric8 Kubernetes client v4.6.1 is compatible with Kubernetes API up to 1.15.2 (refer answer).

You can launch the specific Kubernetes API version with Minikube by adding --kubernetes-version flag to minikube start command (refer docs).

Also the issue might be caused by OkHttp library bug described in the comment of this qustion.

This Answer isnt limited to minikube!

Use a local registry: