# Installation
## Operator
The scenarios related to Installation only cover the targeted version of OpenShift i.e 4.13
### Positive:
- [ ] Kepler operator installs fine from Community Catalog
:::warning
This is to be verified from OpenShift Console
:::
- [ ] Kepler operator installs fine from Operator-SDK.
:::warning
This is to be verified by Upstream/Downstream CI.
:::

- [ ] Kepler operator should be able to deploy/handle any dependencies required.
:::warning
This includes the ability of the operator to provision, configure, and manage resources such as pods, services, volumes, etc.
:::

- [ ] Following resources should get deployed as part of Operator Installation:
- [ ] `kepler-operator-controller-manager deployment`
- [ ] `kube-rbac-proxy`
- [ ] `manager`
- [ ] `kepler-operator-controller-manager replicaset`
- [ ] `kepler-operator-controller-manager-metrics-service`
- [ ] `Kepler CRD`
- [ ] `kepler-operator-controller-manager Service Account`
- [ ] `roles`
- [ ] `rolebindings`
- [ ] `lease`
- [ ] `servicemonitor` -> TBD
- [ ] `kepler-operator CSV`
- [ ] `installplan`
- [ ] `subscription`
- [ ] `packagemanifests`
- [ ] `catalog`
### Negative:
- [ ] OpenShift console should log appropiate error in case of unsucessful installation of Operator.
:::warning
This is to be verified from OpenShift Console.
It can be related to any failure scenario or in-case of unsuccessful installation.
:::

:::spoiler **Out of Dev Preview**
- [ ] Kepler operator installs fine on namespace other than openshift-operators.
:::warning
Try deploying the operator on different namespaces.
:::
- [ ] Kepler Operator should log appropriate status in case of successful installation
:::warning
On the OpenShift Console side we should be able to see the success status
:::
:::

## Kepler Instance
### Positive:
- [ ] Kepler operator when installed from Community Catalog can create a Kepler instance successfully.
:::warning
This is to be verified from OpenShift console
Kepler operator can reconcile `kepler` CR
:::
- [ ] Kepler operator when installed from Operator-SDK can deploy Kepler instance when Kepler template is applied.
:::warning
This is to be verified by Upstream/Downstream CI.
:::
- [ ] Following resources should get deployed as part of Kepler Instance Creation:
- [ ] `openshift-kepler-operator namespace`
- [ ] `clusterrole`
- [ ] `scc`
- [ ] `clusterrolebindings`
- [ ] `kepler-exporter-ds`
- [ ] `kepler-exporter-cm`
- [ ] `kepler-exporter-svc`

- [ ] Kepler operator only exposes `exporter port` at the time of configuring the Kepler Instance.
- [ ] Kepler Instance should be able to run with different port number.
:::warning
This is to be verified from OpenShift console
:::

- [ ] Appropriate status should be reflected for creation of Instance(Reconcillation status).
:::warning
This includes whether Kepler Instance was deployed successfully on cluster nodes and logs any error on controller manager.
:::

:::spoiler **Out of Dev Preview**
- [ ] After Kepler Instance creation Operator should show Instance related objects.
:::warning
This includes what all get deployed as part of Instance creation like daemonsets, services, namespace etc.
:::
- [ ] Kepler exporter daemonsets should only be created on nodes that has label attached to them.
:::warning
If the label is attached to nodes then Kepler instance should respect that else it should stick to the default behavior i.e deploy on all the available nodes.
:::
:::

### Negative:
- [ ] OpenShift/Kubernetes cluster log error if an invalid port value is passed in the Instance configuration.
- [ ] OpenShift/Kubernetes cluster logs an appropriate error if an invalid/wrong configuration key/value is passed.
- [ ] Kepler Operator logs an appropriate error if no nodes are available for deployment of Kepler Instance.
:::warning
Appropriate status should be reflected in this case.
:::
- [ ] Appropriate logging of error if an invalid name of Kepler instance is provided.
:::warning
Kepler-Operator should only respect `kepler` as the instance name. Other than that should be considered invalid.
:::
- [ ] Kepler controller manager logs an appropriate error if Instance creation is unable to create required resources.
:::warning
Availability status should also be updated
:::
- [ ] Kepler controller manager logs an appropriate error if it is unable to pull Kepler Image with a pre-configured tag.
:::warning
Availability status should also be updated
:::

# Metrics
:::spoiler **Kepler Operator Metrics(Out of scope for Dev preview)**
- [ ] Kepler controller manager should be able to expose metrics at the configured port.
- [ ] Metric HTTP endpoint should be reachable either via port-forwarding or creating a route on OpenShift.
- [ ] Following metrics should be available and updated accordingly:
- [ ] `controller_runtime_reconcile_total`
- [ ] `controller_runtime_max_concurrent_reconciles`
- [ ] `leader_election_master_status`
:::

## Kepler Exporter Metrics
- [ ] Kepler exporter should be able to expose metrics at port 9103
- [ ] Metric HTTP endpoint should be reachable either via port-forwarding or creating a route on OpenShift.
- [ ] Following metrics should be available and updated accordingly:
:::info
These metrics are used inside our grafana dashboard.
:::
- [ ] `kepler_container_package_joules_total`
- [ ] `kepler_container_dram_joules_total`
- [ ] `kepler_container_other_host_components_joules_total`
- [ ] `kepler_container_gpu_joules_total`

# Uninstallation
## Kepler Operator
### Positive
- [ ] Kepler Operator should get uninstalled successfully when deployed using OpenShift console.
- [ ] Kepler Operator should get uninstalled successfully when deployed using Operator-SDK.
- [ ] Following resources should get uninstalled:
- [ ] `kepler-operator-controller-manager deployment`
- [ ] `kube-rbac-proxy`
- [ ] `manager`
- [ ] `kepler-operator-controller-manager replicaset`
- [ ] `kepler-operator-controller-manager-metrics-service`
- [ ] `kepler-operator-controller-manager Service Account`
- [ ] `kepler-operator CSV`
- [ ] `subscription`
- [ ] `servicemonitor` -> **TBD**
### Negative
- [ ] Appropriate error must be logged in case of unable to remove Operator from the cluster.
:::warning
Reconcillation should report the appropiate reason.
:::
## Kepler Instance
### Positive
- [ ] Kepler Instance should get uninstalled successfully using OpenShift Console.
:::warning
This is to be verified from OpenShift console
:::
- [ ] Kepler Operator should get uninstalled successfully using Operator-SDK.
:::warning
This is to be verified by Upstream/Downstream CI.
:::
- [ ] Following resources should get uninstalled:
- [ ] `openshift-kepler-operator namespace`
- [ ] `clusterrole`
- [ ] `scc`
- [ ] `clusterrolebindings`
- [ ] `kepler-exporter-ds`
- [ ] `kepler-exporter-cm`
- [ ] `kepler-exporter-svc`
### Negative
- [ ] Controller manager logs an appropriate error if it is unable to remove Kepler Instance from the cluster and availability status should be updated accordingly.
:::warning
If something isn't able to uninstall then the related objects should be present on cluster for debugging purposes.
:::

# Grafana Dashboard
## Positive
- [ ] `deploy-grafana.sh` should be able to deploy/configure the Grafana dashboard.
- [ ] Configuring `user-workload-monitoring` if not already configured.
- [ ] Install Grafana Operator inside `openshift-kepler-operator` namespace.
- [ ] Configure tokens
- [ ] Configure Service monitors
- [ ] Add Grafana dashboard source.
- [ ] Grafana dashboard should not show CO2 carbon emission panels when deployed. (**For Dev Preview**)
- [ ] Grafana dashboard should be able to query data from `user-workload-monitoring` Prometheus.
- [ ] Grafana dashboard should show tooltips with relevant information on each panel.
- [ ] All the panels inside dashboard should update accordingly.

## Negative
- [ ] `deploy-grafana.sh` should log an appropriate error in case of any failure.