Getting Started¶

This guide walks you through accessing SLOpilot for the first time, completing the initial setup, and understanding what to expect during the data collection warm-up period.

1. Accessing the Web Interface¶

After installation, access SLOpilot using one of the following methods depending on your environment:

Port Forwarding (simplest, no ingress required):

kubectl port-forward -n slopilot svc/slopilot-rightsizing 8080:80

Then open http://localhost:8080 in your browser.

Ingress: If you configured ingress during installation (see Configuration), access SLOpilot via the hostname you specified.

NodePort / LoadBalancer: Depending on your service.type configuration, retrieve the address from the post-install notes printed by Helm:

# NodePort
export NODE_PORT=$(kubectl get -n slopilot -o jsonpath="{.spec.ports[0].nodePort}" \
    services slopilot-rightsizing)
export NODE_IP=$(kubectl get nodes -n slopilot \
    -o jsonpath="{.items[0].status.addresses[0].address}")
echo "http://$NODE_IP:$NODE_PORT"

# LoadBalancer
export SERVICE_IP=$(kubectl get svc -n slopilot slopilot-rightsizing \
    --template "{{ range (index .status.loadBalancer.ingress 0) }}{{.}}{{ end }}")
echo "http://$SERVICE_IP"

Default credentials:

Username: admin
Password: provided via --admin-password during installation, or retrieve the auto-generated one:

kubectl get secret slopilot-rightsizing-users -n slopilot \
    -o jsonpath='{.data.default-admin-password}' | base64 -d

On first login you will be prompted to accept the EULA and change the default password. Navigate to Profile (user icon) at any time to update your password. Admins can create additional accounts and assign roles — see Features — User Management for details.

3. Data Collection Warm-Up¶

After deployment, SLOpilot begins collecting metrics from your cluster immediately. Recommendations, however, require a period of data accumulation before they can be generated reliably.

Recommendations are not available immediately

First few days: The workloads table will show an amber Collecting... indicator for all workloads. This is expected behaviour — the analysis engine is gathering the metric history it needs.
After approximately one week: Recommendations begin appearing. Confidence scores will be partial, reflecting the data collected so far.
After approximately two weeks: Recommendations reach full confidence. The analysis engine has a complete picture of your workload patterns across the full observation window.

No action is required during this period. Data collection is fully automatic.

The diagram below shows the data flow from your cluster to a recommendation:

flowchart LR
    A[Kubernetes\nCluster] -->|scrapes metrics| B[Prometheus]
    B -->|queries| C[SLOpilot\nAnalysis Engine]
    A -->|workload\ndiscovery| C
    C -->|confidence\ncheck| D{Sufficient\nData?}
    D -->|Yes| E[Recommendation\nwith Confidence Score]
    D -->|No| F[Collecting...\nAmber Indicator]

4. Navigating the Interface¶

The sidebar provides access to all sections of the application:

Section	Purpose
Dashboard	Cluster overview, top resource consumers, and top optimization opportunities
Namespaces	Browse monitored namespaces and their workload counts
Workloads	Filterable, sortable table of all monitored Deployments and StatefulSets
Reports	Download and manage generated PDF reports
Settings	System status, license information, and Prometheus configuration

For detailed information on each section, see Features.

Getting Started¶

1. Accessing the Web Interface¶

2. First Login¶

3. Data Collection Warm-Up¶

4. Navigating the Interface¶