xhyve | xhyve , a lightweight OS X virtualization solution | Infrastructure Automation library

Your 1.34.0 version of Minishift is currently tested against docker-machine-driver-xhyve version 0.3.3. and has compatibility issues with the recently deployed docker-machine-driver-xhyve v0.4.0.

The solution for me was to uninstall the docker-machine-driver-xhyve 0.4.0 driver and install the old and tested 0.3.3:

https://github.com/machine-drivers/docker-machine-driver-xhyve/releases/tag/v0.3.3

Your question would benefit from your inclusion of the docker-compose.yaml file in its entirety so that we may understand what you're doing.

From what you have included, I have (not mutually exclusive) hypotheses:

Possibility #1: The image builds are probably only run once and not every time you run docker-compose. When you run docker-compose, if it finds the relevant images locally, it won't rebuild them. If you delete the local images, or you force a change, then the images will be rebuilt.

If the images aren't rebuilt, changes to the sources will not be reflected.

Possibility #2: Your Dockerfile uses ADD . /app. When this image is built, the files in your current directory (.) are copied to the image's /app folder. This only occurs during the build.

Possibility #3: You reference volumes and /app but this mount point already exists in the container image for which you included the Dockerfile (ADD . /app). I'm unsure what the consequence of this behavior is but you may be overriding the container's /app directory (which contained the files you ADD . /app). This is redundant.

It is considered to be a not good practice to change source files within a container image. One practice often used with containers is of immutable infrastructure. The idea is that, while the data may change, a container's application|process|binary does not change.

If I were given golang:1.12 today, it should always be exactly the same. If there were a change to Golang 1.12 -- even if it were one variable renamed -- the Go team would upgrade the version and create perhaps Golang 1.12.1. Then I'd expect a new container image golang:1.12.1.

This practice is not enforced by docker tags and this is one (of many) reason(s) why docker tags aren't 'trustworthy'.

The best practice is thus to rebuild an image every time a source file changes.

You will frequently see -- and it's a good mechanism -- that folks will rebuild the container images for every e.g. git commit. The hash of the commit is often used to tag the container image too but this is optional.

Try Mac's pfctl command, it's kind of equivalent to iptables.

Here's man page: https://developer.apple.com/legacy/library/documentation/Darwin/Reference/ManPages/man8/pfctl.8.html

I am not sure what was the problem. But I uninstalled my docker and installed it again with the dmg file provided by docker. It works now and I am finally able to access elasticsearch :)

On GKE or AWS installs, the external IP comes from the cloud support that reports back to kube API the address that the created LB was assigned.

To have the same on minikube you'd have to run some kind of an LB controller, ie. haproxy one, but honestly, for minikube it makes little sense, as you have single IP that you know in advance by minikube ip so you can use NodePort with that knowledge. LB solution would require setting some IP rangethat can be mapped to particular nodeports, as this is effectively what LB will do - take traffic from extIP:extPort and proxy it to minikubeIP:NodePort.

Unless your use case prevents you from it, you should consider Ingress as the way of ingesting traffic to your minikube.

Upgrading minishift to release 1.5 fixed the issue.

If you are using minikube addons enable ingress then an ingress controller would be already deployed (not being necessary to deploy your own). Maybe there is a conflict with the two ingress controllers. My advice would be to remove the controller you created and stick to the already deployed by the minikube ingress addon.

To avoid re-building the docker image you can transfer an image built on your local machine into the minikube VM with:

I found the answer. Apparently VBoxManage is picky about what user it uses. I was able to get it to work with sudo -u USERNAME minikube ip