3.1   Requirements¶

We assume a Linux box running Centos 7. We’ve installed Docker CE and kubectl:

• Docker CE 18.09.6
• kubectl 1.14.1

3.2   Configure Docker to start on boot¶

CentOS uses systemd to manage which services start when the system boots. Run the following to configure Docker to start on boot.

sudo systemctl enable docker

3.3   Start the k3s master¶

Start the k3s master with the following commands:

export K3S_PORT=6443
export K3S_URL=http://localhost:${K3S_PORT}
export K3S_TOKEN=$(date | base64)
export HOST_PATH=/data # change depending on your host
export CONTAINER_PATH=/opt/local-path-provisioner
sudo docker run  -d --restart always --tmpfs /run --tmpfs /var/run --volume ${HOST_PATH}:${CONTAINER_PATH} -e K3S_URL=${K3S_URL} -e K3S_TOKEN=${K3S_TOKEN} --privileged --network host --name master docker.io/rancher/k3s:v0.5.0-rc1 server --https-listen-port ${K3S_PORT} --no-deploy traefik

The --restart always option ensures that the k3s master is automatically restarted after a system reboot.

Data is persisted at $HOST_PATH with the --volume option (see also 3.4   Deploy the local-path provisioner).

With the --network host option, network traffic is routed from the host to the container, we need that in order to reach the different services running on k3s.

Note that we are not deploying Traefik Ingress Controller which is included in the k3s docker image, because the DM-EFD already deploys the NGINX Ingress Controller.

To connect to the master you need to copy the kubeconfig file from the container, and set the KUBECONFIG environment variable so that kubectl knows how to connect to the cluster:

sudo docker cp master:/etc/rancher/k3s/k3s.yaml k3s.yaml
export KUBECONFIG=$(pwd)/k3s.yaml
kubectl cluster-info

Kubernetes master is running at https://localhost:6443
CoreDNS is running at https://localhost:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

To connect to the cluster from an another machine, copy the k3s.yaml file and replace localhost by 140.252.32.142.

3.4   Deploy the local-path provisioner¶

The local-path provisioner will create hostPath persistent volumes on the node automatically. The directory /opt/local-path-provisioner will be used as the path for provisioning. The provisioner will be installed in the local-path-storage namespace by default.

kubectl apply -f https://raw.githubusercontent.com/rancher/local-path-provisioner/master/deploy/local-path-storage.yaml

At this point you should see the following pods running in the cluster:

kubectl get pods --all-namespaces
NAMESPACE            NAME                                      READY   STATUS    RESTARTS   AGE
kube-system          coredns-695688789-r9gkt                   1/1     Running   0          5m
local-path-storage   local-path-provisioner-5d4b898474-vz2np   1/1     Running   0          4s

3.5   Add workers (optional)¶

If there are more machines, you can easily add workers to the cluster. Copy the node-token from the master:

sudo docker cp master:/var/lib/rancher/k3s/server/node-token node-token

and start the worker(s):

export SERVER_URL=https://<master external IP>:${K3S_PORT}
export NODE_TOKEN=$(cat node-token)
export WORKER=kube-0
export HOST_PATH=/data # change depending on your host
export CONTAINER_PATH=/opt/local-path-provisioner
sudo docker run -d --tmpfs /run --tmpfs /var/run -v ${HOST_PATH}:${CONTAINER_PATH} -e K3S_URL=${SERVER_URL} -e K3S_TOKEN=${NODE_TOKEN} --privileged --name ${WORKER} rancher/k3s:v0.5.0-rc1

4.1   Requirements¶

• AWS credentials (we save the deployment configuration to an S3 bucket and create names for our services on Route53)
• TLS/SSL certificates for the lsst.codes domain (certificates are shared via SQuaRE Dropbox account)
• Deployment configuration for the DM-EFD test environment (secrets are shared via SQuaRE 1Password account)

We automate the deployment of the DM-EFD with Terraform and Helm. Terragrunt is used to manage the different deployment environments (dev, test, stage, and production) while keeping the Terraform modules environment-agnostic. We also use Terragrunt to save the Terraform configuration and the state of a particular deployment remotely.

Install Terragrunt, Terraform, and Helm.

git clone https://github.com/lsst-sqre/terragrunt-live-test.git
cd terragrunt-live-test
make all
export PATH="${PWD}/bin:${PATH}"

Install the SSL certificates (this step requires access to the SQuaRE Dropbox account).

make tls

4.2   Initialize the deployment environment¶

The following commands initialize the deployment environment. (Terragrunt uses an S3 bucket to save the deployment configuration, so this step requires the AWS credentials).

export AWS_ACCESS_KEY_ID=""
export AWS_SECRET_ACCESS_KEY=""

cd afausti/efd
make all
terragrunt init --terragrunt-source-update
terragrunt init

4.3   Deployment configuration¶

The DM-EFD deployment configuration on k3s is defined by a set of Terraform variables listed in the terraform-efd-k3s repository.

Edit the terraform.tfvars file with the values obtained from the SQuaRE 1Password account. Search for terraform vars (efd test).

Finally deploy the DM-EFD with the following commands:

terragrunt plan
terragrunt apply

4.4   Outputs¶

If everything is correct you should see an output similar to this, indicating the services deployed:

confluent_lb_ips = [140.252.32.142]
grafana_fqdn = test-grafana-efd.lsst.codes
influxdb_fqdn = test-influxdb-efd.lsst.codes
nginx_ingress_ip = 140.252.32.142
prometheus_fqdn = test-prometheus-efd.lsst.codes

The Kafka cluster can be reached at test-efd.lsst.codes:31090.

7.1   Initialize a SAL subsystem¶

The following command will initialize a SAL subsystem, deploy the corresponding SAL transform app and configure the InfluxDB Sink Connector to consume the SAL topics of that subsystem. In the example, the ATCamera:

helm install --name ATCamera --namespace kafka-efd-apps --set subsystem=ATCamera lsstsqre/kafka-efd-apps

7.2   Check a SAL transform app¶

Inspect the logs of a SAL transform app for a particular subsystem. In this example the ATCamera:

kubectl logs $(kubectl get pods --namespace kafka-efd-apps -l "app=saltransform,subsystem=ATCamera" -o jsonpath="{.items[0].metadata.name}") --n kafka-efd-apps

7.3   Check the InfluxDB Sink connector¶

Inspecting the Kafka connect logs:

kubectl logs $(kubectl get pods --namespace kafka -l "app=cp-kafka-connect,release=confluent" -o jsonpath="{.items[0].metadata.name}") cp-kafka-connect-server --namespace kafka -f

7.4   Getting data from the DM-EFD¶

InfluxDB provides an HTTP API for accessing the data. A Python code snippet to get data from a particular SAL topic from the DM-EFD is shown below. In this example, we retrieve the Temperature for CCD 0 in the last 24h:

import requests

INFLUXDB_API_URL = "https://test-influxdb-efd.lsst.codes"
INFLUXDB_DATABASE = "efd"

def get_topic_data(topic, period="24h"):
  params={'q': 'SELECT * FROM "{}\"."autogen"."{}" where time > now()-{}'.format(INFLUXDB_DATABASE, topic, period)}
  r = requests.post(url=INFLUXDB_API_URL + "/query", params=params)

  return r.json()

get_topic_data("lsst.sal.ATCamera.ccdTemp0")