Camel K selective upgrade demo

This demo shows how two Camel K operators can be set to run in parallel and reconcile their own integrations, making it possible to selectively upgrade and downgrade individual integrations across the cluster.

Setup

This is a POC and runs on Minikube. It requires Camel K 1.7.0 (unreleased at the time of writing this demo) or Camel K 1.7.0-SNAPSHOT compiled from the release-1.7.x (or main until release) branch.

Using:

• kamel Camel K CLI (see above for version version)
• kubectl tool
• stern tool to display the logs (or similar)

1. Prepare the cluster

The simplest way to prepare roles and service accounts for the two operators is to install Camel K using the CLI:

kamel install --global -n default -w --force

We don't need the default operator, so we remove the deployment and platform related resources:

kubectl delete deployment camel-k-operator
kubectl delete ip --all
kubectl delete cm camel-k-maven-settings || true

Now we can install the two operators.

NOTE: you must edit the setup/platform-* resources to set your own ip address of the Minikube registry. You need also set the proper operator image if you're not using a SNAPSHOT.

After doing the changes, to install the two operators:

kubectl apply -f setup/

You should see two operators in the default namespace and both integration platforms should be quickly reconciled.

2. Run the connectors

Create a user namespace and switch to it:

kubectl create ns user
kubectl config set-context --current --namespace user

Now we can create two connectors associated to the first operator:

kubectl apply -f my-connector-1.yaml
kubectl apply -f my-connector-2.yaml

When kubectl get klb returns that they are Running, it means that the first operator reconciled both of them, while the second is idle (and printing messages indicating that it's ignoring those resources).

3. Watching the two connectors

The two connectors just print messages to the logs. To distinguish the version of the operator they are running with, you can notice:

• The v1 operator/platform will materialize integrations that don't use colors in the logs
• The v2 operator/platform will materialize integrations that do use colors in the logs

You can see the logs using (in another terminal):

stern my-connector

Both integrations will have uncolored logs.

4. Upgrading integrations

The script upgrade.sh can be used to upgrade an integration to v2, while downgrade.sh will downgrade it to v1. Upgrading and downgrading mean passing the integration management responsibility from one operator to the other.

You can play with the connectors:

./upgrade.sh my-connector-1
sleep 30
./upgrade.sh my-connector-2
sleep 30
./downgrade.sh my-connector-1
./downgrade.sh my-connector-2

What's inside the upgrade and downgrade scripts can be automated in a "supervisor" operator.