Indicate clearly if we expect people to deploy
Prometheus or not. Explain better what the Helm
deployment does. Add a conclusion slide about
Grafana dashboards.
Prometheus deployment with Helm now stores
correctly Helm files in ~docker instead of
~ubuntu.
This will be used for kubernetes admin labs, to upgrade
an existing cluster. In order to be able to perform an
upgrade, we need a cluster running an older version.
In some cases, I would like Prometheus to be pre-installed (so that
it shows a bunch of metrics) without relying on people doing it (and
setting up Helm correctly). This patch allows to run:
./workshopctl helmprom TAG
It will setup Helm with a proper service account, then deploy
the Pormetheus chart, disabling the alert manager, persistence,
and assigning the Prometheus server to NodePort 30090.
This command is idempotent.
This adds a few features:
- ./workshopctl kubereset TAG (closes#306)
- remove python-setuptools (prepare for #353)
- ./workshopctl weavetest TAG (help detecting weave issues
like we had at OSCON, July 2018)
- remove a bit of dead code
kubectl and kubens are added as kctl and kns (to avoid clashing with
completion for kubectl). Their completion is added too (so you can
do 'kns kube-sy[TAB]' to switch to kube-system).
kube_ps1 is added and enabled. The default prompt for the docker
user now shows the current context and namespace.
This allows to manage groups of VMs across multiple infrastructure
providers. It also adds support to create groups of VMs on OpenStack.
WARNING: the syntax of workshopctl has changed slightly. Check READMEs
for details.
The commands to install Stern and Helm aren't super exciting,
so let's pre-install these tools. That way, we also generate
completion for them. We still give installation instructions
just in case, but this saves time for more important stuff.
The kubetest command used to say [SUCCESS] on completely
fresh nodes. Now we check the existence of the /tmp/node
file, as well as of the kubectl executable.
`trap ... ERR` does not automatically propagate to functions. Therefore,
Our fancy error-reporting mechanism did not catch errors happening in
functions; and we do most of the actual work in functions. The solution
is to `set -E` or `set -o errtrace`.
