===== Overview =====

How to install Kubernetes on Fedora 28 using downloadable binary tarballs, based on the installation script you can find [[https://get.k8s.io/|here]]. What this page does is explain what that script accomplishes. Note this does not allow you to build the source, only to run the binaries.

===== The tarballs =====

The ones you need:
  * [[https://dl.k8s.io/v1.10.0/kubernetes.tar.gz|kubernetes.tar.gz]]
  * [[https://dl.k8s.io/v1.10.0/kubernetes-client-linux-amd64.tar.gz|kubernetes-client-linux-amd64.tar.gz]]
  * [[https://dl.k8s.io/v1.10.0/kubernetes-server-linux-amd64.tar.gz|kubernetes-server-linux-amd64.tar.gz]]

The rest (not sure you need these):

  * [[https://dl.k8s.io/v1.10.0/kubernetes-src.tar.gz|kubernetes-src.tar.gz]]
  * [[https://dl.k8s.io/v1.10.0/kubernetes-node-linux-amd64.tar.gz|kubernetes-node-linux-amd64.tar.gz]]

===== kubernetes.tar.gz =====

This tarball kicks things off, as the installation script ends with:

<code>
echo "Unpacking kubernetes release ${KUBE_VERSION}"
rm -rf "${PWD}/kubernetes"
tar -xzf ${file}

download_kube_binaries
create_cluster
</code>

So unload that tarball and get to work.



===== download_kube_binaries =====

==== Overview ====

This function call does two things:

  * downloads the server tarball, but leaves it unextracted
  * downloads //and extracts// the client tarball

==== download_kube_binaries() ====

This just turns around and calls a script for the download and extraction:

<code>
# Use the script from inside the Kubernetes tarball to fetch the client and
# server binaries (if not included in kubernetes.tar.gz).
function download_kube_binaries {
  (
    cd kubernetes
    if [[ -x ./cluster/get-kube-binaries.sh ]]; then
      # Make sure to use the same download URL in get-kube-binaries.sh
      KUBERNETES_RELEASE_URL="${KUBERNETES_RELEASE_URL}" \
        ./cluster/get-kube-binaries.sh
    fi
  )
}
</code>

==== cluster/get-kube-binaries.sh ====

=== kubernetes-server-linux-amd64.tar.gz ===

As you can see, the server tarball is downloaded and simply left under ''server/'' for later:

<code>
if "${DOWNLOAD_SERVER_TAR}"; then
  download_tarball "${KUBE_ROOT}/server" "${SERVER_TAR}"
fi
</code>

so you can easily do that manually.

=== kubernetes-client-linux-amd64.tar.gz ===

Unlike the server tarball, the client tarball is not only downloaded under ''client/'', it is then "extracted":

<code>
if "${DOWNLOAD_CLIENT_TAR}"; then
  download_tarball "${KUBE_ROOT}/client" "${CLIENT_TAR}"
  extract_arch_tarball "${KUBE_ROOT}/client/${CLIENT_TAR}" "${CLIENT_PLATFORM}" "${CLIENT_ARCH}"
fi
</code>

The client tarball consists of the single executable ''kubectl'':

<code>
$ tar tvf kubernetes-client-linux-amd64.tar.gz 
drwxr-xr-x root/root         0 2018-03-26 13:46 kubernetes/
drwxr-xr-x root/root         0 2018-03-26 13:46 kubernetes/client/
drwxr-xr-x root/root         0 2018-03-26 13:46 kubernetes/client/bin/
-rwxr-xr-x root/root  54277699 2018-03-26 13:46 kubernetes/client/bin/kubectl
$
</code>

which will be "installed" and which you'll have to add to your search path shortly.

Here's the ''extract_arch_tarball()'' function in that script:

<code>
function extract_arch_tarball() {
  local -r tarfile="$1"
  local -r platform="$2"
  local -r arch="$3"

  platforms_dir="${KUBE_ROOT}/platforms/${platform}/${arch}"
  echo "Extracting ${tarfile} into ${platforms_dir}"
  mkdir -p "${platforms_dir}"
  # Tarball looks like kubernetes/{client,server}/bin/BINARY"
  tar -xzf "${tarfile}" --strip-components 3 -C "${platforms_dir}"
  # Create convenience symlink
  ln -sf "${platforms_dir}" "$(dirname ${tarfile})/bin"
  echo "Add '$(dirname ${tarfile})/bin' to your PATH to use newly-installed binaries."
}
</code>

In short, from the ''${KUBE_ROOT}'' directory (and in the context of a 64-bit Fedora system), extract that single ''kubectl'' executable:

<code>
$ mkdir -p platforms/linux/amd64
$ tar xvf ../kubernetes-client-linux-amd64.tar.gz \
> --strip-components 3 \
> -C platforms/linux/amd64
$
</code>

<code>
$ tree platforms
platforms
└── linux
    └── amd64
        └── kubectl
$
</code>

Finally, make sure you can find ''kubectl'' on your search path somehow:

<code>
PATH=$PATH:/home/rpjday/kubernetes/dl/kubernetes/platforms/linux/amd64
</code>
===== create_cluster =====

==== Overview ====

This function kicks off a new cluster:

<code>
function create_cluster {
  if [[ -n "${KUBERNETES_SKIP_CREATE_CLUSTER-}" ]]; then
    exit 0
  fi
  echo "Creating a kubernetes on ${KUBERNETES_PROVIDER:-gce}..."
  (
    cd kubernetes
    ./cluster/kube-up.sh
    echo "Kubernetes binaries at ${PWD}/cluster/"
    if [[ ":$PATH:" != *":${PWD}/cluster:"* ]]; then
      echo "You may want to add this directory to your PATH in \$HOME/.profile"
    fi

    echo "Installation successful!"
  )
}
</code>

You might want to add the ''cluster/'' directory to your search path for later:

<code>
PATH=$PATH:/home/rpjday/kubernetes/dl/kubernetes/cluster
</code>
==== create_cluster() ====

This function calls the underlying script ''cluster/kube-up.sh'':

<code>
function create_cluster {
  if [[ -n "${KUBERNETES_SKIP_CREATE_CLUSTER-}" ]]; then
    exit 0
  fi
  echo "Creating a kubernetes on ${KUBERNETES_PROVIDER:-gce}..."
  (
    cd kubernetes
    ./cluster/kube-up.sh
    echo "Kubernetes binaries at ${PWD}/cluster/"
    if [[ ":$PATH:" != *":${PWD}/cluster:"* ]]; then
      echo "You may want to add this directory to your PATH in \$HOME/.profile"
    fi

    echo "Installation successful!"
  )
}
</code>