Pre-requisites

Minimum requirements

• An Ubuntu 20.04 LTS, 18.04 LTS or 16.04 LTS environment to run the commands
• At least 20G of disk space and 8G of memory are recommended
• An internet connection

Prepare environment

Microk8s:

sudo snap install microk8s --classic --channel 1.19/stable

sudo usermod -a -G microk8s `whoami`
sudo chown -f -R `whoami` ~/.kube

newgrp microk8s
microk8s.status --wait-ready
microk8s.enable storage dns

Juju controller:

sudo snap install juju --classic --channel 2.8/stable
juju bootstrap microk8s

Juju model:

juju add-model ecosystem microk8s

Build charms:

./build_charms.sh

Requirement for elasticsearch:

sudo sysctl -w vm.max_map_count=262144

Deployment

To deploy the whole environment at the same time, execute the following command:

juju deploy cs:knf-lma-stack

The following sections show the steps to deploy the charms manually, one by one.

Monitoring

Prometheus:

juju deploy ./prometheus

Grafana:

juju deploy ./grafana

Relate prometheus and grafana:

juju relate grafana prometheus

Deploy simulator:

juju deploy ./mock-knf

Relate simulator to prometheus:

juju relate prometheus mock-knf

Scale simulator:

juju scale-application mock-knf 3

Test

Go to Prometheus url and check under status/targets if all node exporters are listed there.

Login to grafana, create a dashboard and check the metrics for the following queries:

rate(node_cpu_seconds_total{instance="mock-knf-0.mock-knf-endpoints:9100"}[5m])
rate(node_cpu_seconds_total{instance="mock-knf-1.mock-knf-endpoints:9100"}[5m])
rate(node_cpu_seconds_total{instance="mock-knf-1.mock-knf-endpoints:9100"}[5m])

Logging

Deploy Graylog:

juju deploy ./graylog
juju config graylog admin-password=admin

Deploy Mongo and elasticsearch:

juju deploy ./mongodb
juju deploy ./elasticsearch

Graylog relations:

juju relate graylog mongodb
juju relate graylog elasticsearch

Deploy Filebeat and relate to graylog:

juju deploy ./filebeat
juju relate graylog filebeat

Scaling

juju scale-application mock-knf 3

Known issues

MongoDB takes a lot of time to configure the replica set

In the start hook, MongoDB will try to initialize the replica set. Normally, when Juju will trigger that hook, the mongod service of the workload pod won't be ready, so the start hook will be deferred. This mean that before running the next hook, the start hook will be executed again. This process keeps going until the mongod service is up and running, and the charm can configure the replica set.

Usually, the update-status hook will make possible the re-trigger of the start hook in order to configure the replica set. The update-status-hook-interval defaults to 5m. If we want to have a faster deployment, we can override the value update-status-hook-interval to 1m.

juju model-config update-status-hook-interval=1m