This repository contains the necessary scripts and templates to deploy the SolarPeformanceInsight on a fresh AWS EKS cluster. View cluster and app metrics at https://grafana.solarperformanceinsight.org and see the status of app deployments at https://argocd.solarperformanceinsight.org by logging in via GitHub.
In the cluster-setup/ folder, you'll find the cluster.yaml file that
was used with eksctl create cluster -f cluster.yaml to create the
initial AWS EKS cluster. One may need to use eksctl utils write-kubeconfig with appropriate options to make sure kubectl
can communicate with the new cluster. Once the cluster is created,
various infrastructure components are installed via Ansible:
ansible-playbook ansible/setup.yml.
After this is complete (and routes are created to the NLB in Route53)
one should be able to login via GitHub to view metrics in Grafana
and see the status of app deployments in Argo CD.
Access to Prometheus
and Alertmanager
will require the addition of htpasswd lines in the
/k8ssecrets/default-htpasswd parameter in the SPI AWS Systems
Manager Parameter Store.
The actual SolarPerformanceInsight app is deployed via Argo
CD from the applications
defined in the applications folder.
Each application should reference a folder containing a kustomization overlay
on top of resources found in base. For example, the development application
builds the Kubernetes manifests from the overlays/development folder. Any
changes to the application should be done through these manifests to let Argo CD
automatically update the resources once the change is merged. Otherwise,
Argo will likely revert any changes made to the resources to the state describe
in the latest commit.
Application errors are monitored with Sentry. Other application statistics can be found on the appropriate dashboard of the Grafana deployment that monitors Prometheus metrics for the cluster. Alerts can be configured either directly in Grafana or through Alertmanager. Application logs are aggregated via Loki and accessible from the Explore link in Grafana.
The production configuration is found in overlays/production.
To deploy a new version of SPI to production, first modify the image hash
in overlays/production/spi/kustomization.yaml. Then, tag the commit with
the change with v0 to have Argo deploy it. To change the tag that deploys
production, edit the targetRevision in applications/production.yaml.