Starship Helm charts

中文

Helm charts for deploying Starship， Tricorder Observability's next-generation Observability platform.

🖖 Starship 🪐 is a next-generation Observability platform built on 🐝 eBPF➕WASM

🚀 Starship is to modern Observability, as ChatGPT is to consumer knownledge discovery. 🐝 eBPF enables instrumentation-free data collection, and WASM complements eBPF's inability to perform complex data processing.

Starship currently only runs on Kubernetes. Starship provides eBPF-powered instrumentation-free Service Map. No need to change a single line of code in your application, instantly access a single-pane view of the high-level status of your Cloud Native applications on Kubernetes.

Starship also collects data from Prometheus and OpenTelemetry.

Starship uses Grafana for visualization. Use the following info to logon Grafana:

username: admin
password: tricorder

Prerequisites

TODO: Add instructions for other public Clouds.

AWS EKS

• If you are using AWS EKS, install EBS CSI on your EKS cluster. This is required because Helm charts create PersistentVolume for database pods, which requires EBS CSI.

ALIYUN ACK

• If you are using Aliyun ACK, default install Aliyun CSI on your ACK cluster. This is required because Helm charts create PersistentVolume for database pods, which requires Aliyun CSI.
• Before installing Starship, you need to check default storageclass,

  kubectl get storageclass | grep default

• If your cluster has no default storageclass, you'll need to run the command below to create the default storageclass

  kubectl patch storageclass <you-storageclass-name> --patch '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'
  # example:
  kubectl patch storageclass alibabacloud-cnfs-nas --patch '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'

  WARNING: Starship installation will fail if there is no default storageclass on your cluster.

Install

Change namespace to your own, here we use tricorder as an example.

helm repo add tricorder-stable \
    https://tricorder-observability.github.io/helm-charts
helm repo update
kubectl create namespace tricorder
helm install my-starship tricorder-stable/starship -n tricorder

Access Starship web UI through LoadBalancer external IP

Starship by default expose web UI service through LoadBalancer service. If your cluster has configured LoadBalancer that supports external access, like AWS LoadBalancer Controller or an Ingress Controller, you can access the service directly through api-server service's ExteranIP:

kubectl get service -n tricorder

Navigate to http://${EXTERNAL-IP} in your browser to access Starship's Web UI, note that the protocol is HTTP, not HTTPS. You will be able to open Grafana instance by following the link on the left panel of the management Web UI.

Access Starship Web UI through kubectl port-forward

You can expose the services to your local network using the kubectl port-forward command. Use the following command to expose the Starship managenment UI service on port 18080 on localhost: http://localhost:18080.

kubectl -n tricorder port-forward service/api-server 18080:80

Access Starship Web UI through Ingress

Ingress exposes HTTP and HTTPS routes from outside the cluster to services with in the cluster. Use the following command to expose the Starship managenment UI service as tstarship.io host on ingress port 80.

kubectl apply -f - <<EOF
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: tricorder
  namespace: tricorder
spec:
  rules:
  - host: starship.io
    http:
      paths:
      - pathType: Prefix
        path: "/"
        backend:
          service:
            name: api-server
            port:
              number: 80
EOF

If you want to use more ingress features, such as TLS termination, please follow this documentation to write ingress configuration.

Caveats

WARNING Do not install multiple Starship releases in multiple namespaces. That won't work because of system limitations. If you accidentally did that, follow the uninstall instructions to remove all artifacts and reinstall.

WARNING Do not install multiple releases in the same namespace.

WARNING: This project is currently in active development. Consider this a technical preview only.

Data Retention

Metric and Trace data has an automated retention that drops data after a certain age. The default retention is 7 days:

promscale:
  config:
    startup.dataset.config: |
      metrics:
        compress_data: true
        default_retention_period: 7d
      traces:
        default_retention_period: 7d

and above retention can be customized by --values flag, We can change default_retention_period's value from 7 days to 30 days:

• create patch yaml file for custom values:

cat > rentention_patch.yaml << EOF
promscale:
  config:
    startup.dataset.config: |
      metrics:
        compress_data: true
        default_retention_period: 30d
      traces:
        default_retention_period: 30d
EOF

• override default settings by --values flag:

helm install my-starship tricorder-stable/starship -n tricorder \
    --values rentention_patch.yaml

Uninstall

To uninstall a release you can run:

helm uninstall my-starship -n tricorder

After uninstalling helm release some objects will be left over. To remove them follow next sections.

Cleanup secrets

Secret's created with the deployment aren't deleted. These secrets need to be manually deleted:

kubectl delete -n tricorder \
    $(kubectl get secrets -n tricorder -l "app=timescaledb" -o name)

Cleanup configmap

kubectl delete -n tricorder \
    $(kubectl get configmap -n tricorder -l "app=my-starship-promscale" -o name)

Cleanup Kube-Prometheus secret

One of the Kube-Prometheus secrets created with the deployment isn't deleted. This secret needs to be manually deleted:

kubectl delete secret -n tricorder my-starship-kube-prometheus-stack-admission

Cleanup DB PVCs and Backup

Removing the deployment does not remove the Persistent Volume Claims (pvc) belonging to the release. For a full cleanup run:

kubectl delete -n tricorder \
    $(kubectl get pvc -n tricorder -l release=my-starship -o name)

Prometheus PVCs

Removing the deployment does not remove the Persistent Volume Claims (pvc) of Prometheus belonging to the release. For a full cleanup run:

kubectl delete -n tricorder $(kubectl get pvc -n tricorder \
  -l operator.prometheus.io/name=my-starship-kube-prometheus-stack-prometheus \
  -o name)

Prometheus CRDs, ValidatingWebhookConfiguration and MutatingWebhookConfiguration

kubectl delete crd alertmanagerconfigs.monitoring.coreos.com \
    alertmanagers.monitoring.coreos.com \
    probes.monitoring.coreos.com \
    prometheuses.monitoring.coreos.com \
    prometheusrules.monitoring.coreos.com \
    servicemonitors.monitoring.coreos.com \
    thanosrulers.monitoring.coreos.com \
    podmonitors.monitoring.coreos.com

kubectl delete MutatingWebhookConfiguration my-starship-kube-promethe-admission
kubectl delete ValidatingWebhookConfiguration my-starship-kube-prometheus-admission

Delete Namespace

kubectl delete namespace tricorder

Advanced topics

Send OpenTelemetry data to Starship

Send OpenTelemetry data to Starship.

Override default values

You can override configuration values defined in charts/starship/Values.yaml with --set flags.

# Override Starship service's type to ClusterIP
helm upgrade my-starship tricorder-stable/starship -n tricorder \
    --set service.type=ClusterIP

# Override Starship container images' tag
helm upgrade my-starship tricorder-stable/starship -n tricorder \
    --set images.tag=<a specific tag>

# Override Starship container images' registry
helm upgrade my-starship tricorder-stable/starship -n tricorder \
    --set images.registry=<a specific imageRegistry>

# Override Starship api-server container image's tag
helm upgrade my-starship tricorder-stable/starship -n tricorder \
    --set apiServer.image.tag=<a specific tag>

Install from local helm-charts repo

git clone this repo, and replace tricorder-stable/starship with the charts local path charts/starship.

git clone git@github.com:tricorder-observability/helm-charts.git
cd helm-charts
# You'll need this step to fetch the dependent charts
helm dep update charts/starship
helm install my-starship charts/starship -n tricorder

All commands listed in the previous install section works when you swap tricorder-stable/starship with charts/starship when the PWD is the root of the repo.