Deploying to GCP

First, we need to set up a Kubernetes cluster (see the official GCP documentation for more info).

Follow the steps on this page to install SigNoz on Kubernetes with Helm.

The SigNoz Helm chart will install the following components into your Kubernetes cluster:

• Query Service (backend service)
• Web UI (frontend)
• OpenTelemetry Collectors
• Alertmanager
• ClickHouse chart (datastore)
• K8s-Infra chart (k8s infra metrics/logs collectors)

Prerequisites

• You must have a GKE cluster. Both Standard and Autopilot are supported.

• Kubernetes version >= 1.22

• x86-64/amd64 workloads as currently arm64 architecture is not supported

• Helm version >= 3.8

• You must have kubectl access to your cluster

• The following table describes the hardware requirements that are needed to install SigNoz on Kubernetes:

  Component	Minimal Requirements	Recommended
  Memory	8 GB	16 GB
  CPU	4 cores	8 cores
  Storage	30 GB	80 GB

• Suggestion: In case you want to use your own custom storage class for PVCs, you can set global.storageClass configuration to desired storage class.

Chart configuration

Here's the minimal required override-values.yaml that we'll be using later. You can find an overview of the parameters that can be configured during installation under chart configuration.

GKE Standard

In GKE Standard, you can either install with the default configuration or make use of the following override-values.yaml:

global:
  storageClass: gce-resizable
  cloud: gcp

clickhouse:
  installCustomStorageClass: true

GKE Autopilot

In GKE Autopilot, you must set cloud to gcp/autogke as well as update kubeletMetrics to use read-only Kubelet endpoint as shown below in the override-values.yaml:

global:
  storageClass: gce-resizable
  cloud: gcp/autogke

clickhouse:
  installCustomStorageClass: true

k8s-infra:
  presets:
    kubeletMetrics:
      authType: none
      endpoint: ${K8S_NODE_NAME}:10255

GKE Autopilot automatically overriddes resource requests/limits. In our case, all signoz chart components as well as components from clickhouse and k8s-infra charts, if enabled. Therefore, make sure to have enough resource quota for the region where the cluster is deployed. Read more about it here.

Install SigNoz on Kubernetes with Helm

1. Add the SigNoz Helm repository to your client with name signoz by running the following command:

   helm repo add signoz https://charts.signoz.io
   

2. Verify that the repository is accessible to the Helm CLI by entering the following command:

   helm repo list
   

3. Use the kubectl create ns command to create a new namespace. SigNoz recommends you use platform for your new namespace:

   kubectl create ns platform
   

4. Run the following command to install the chart with the release name my-release and namespace platform:

helm --namespace platform install my-release signoz/signoz -f override-values.yaml

Output:

NAME: my-release
LAST DEPLOYED: Mon May 23 20:34:55 2022
NAMESPACE: platform
STATUS: deployed
REVISION: 1
NOTES:
1. You have just deployed SigNoz cluster:

- frontend version: '0.8.0'
- query-service version: '0.8.0'
- alertmanager version: '0.23.0-0.1'
- otel-collector version: '0.43.0-0.1'
- otel-collector-metrics version: '0.43.0-0.1'

*Note that the above command installs the latest stable version of SigNoz.

(Optional) To install a different version, you can use the --set flag to specify the version you wish to install. The following example command installs SigNoz version 0.8.0:

helm --namespace platform install my-release signoz/signoz \
  --set frontend.image.tag="0.8.0" \
  --set queryService.image.tag="0.8.0"

5. You can access SigNoz by setting up port forwarding and browsing to the specified port. The following kubectl port-forward example command forwards all connections made to localhost:3301 to <signoz-frontend-service>:3301:

export SERVICE_NAME=$(kubectl get svc --namespace platform -l "app.kubernetes.io/component=frontend" -o jsonpath="{.items[0].metadata.name}")

kubectl --namespace platform port-forward svc/$SERVICE_NAME 3301:3301

Verify the Installation

Using the kubectl -n platform get pods command, monitor the SigNoz deployment process. Wait for all the pods to be in running state:

kubectl -n platform get pods

Output:

NAME                                                        READY   STATUS    RESTARTS   AGE
chi-signoz-cluster-0-0-0                                    1/1     Running   0          8m21s
clickhouse-operator-8cff468-n5s99                           2/2     Running   0          8m55s
my-release-signoz-alertmanager-0                            1/1     Running   0          8m54s
my-release-signoz-frontend-78774f44d7-wl87p                 1/1     Running   0          8m55s
my-release-signoz-otel-collector-66c8c7dc9d-d8v5c           1/1     Running   0          8m55s
my-release-signoz-otel-collector-metrics-68bcfd5556-9tkgh   1/1     Running   0          8m55s
my-release-signoz-query-service-0                           1/1     Running   0          8m54s
my-release-zookeeper-0                                      1/1     Running   0          8m54s

(Optional) Install a Sample Application and Generate Tracing Data

Follow the steps in this section to install a sample application named HotR.O.D, and generate tracing data.

1. Use the HotROD install script below to create a sample-application namespace and deploy HotROD application on it:

   curl -sL https://github.com/SigNoz/signoz/raw/develop/sample-apps/hotrod/hotrod-install.sh \
     | HELM_RELEASE=my-release SIGNOZ_NAMESPACE=platform bash
   

2. Using the kubectl -n sample-application get pods command, monitor the sample application pods. Wait for all the pods to be in running state:

   kubectl -n sample-application get pods
   

   Output:

   NAME                            READY   STATUS    RESTARTS   AGE
   hotrod-55bd58cc8d-mzxq8         1/1     Running   0          2m
   locust-master-b65744bbf-l7v7n   1/1     Running   0          2m
   locust-slave-688c86bcb7-ngx7w   1/1     Running   0          2m
   

3. Use the following command to generate load:

   kubectl --namespace sample-application run strzal --image=djbingham/curl \
     --restart='OnFailure' -i --tty --rm --command -- curl -X POST -F \
     'user_count=6' -F 'spawn_rate=2' http://locust-master:8089/swarm
   

4. Browse to http://localhost:3301 and see the metrics and traces for your sample application.

1. Use the following command to stop load generation:

   kubectl -n sample-application run strzal --image=djbingham/curl \
     --restart='OnFailure' -i --tty --rm --command -- curl \
     http://locust-master:8089/stop
   

Go to Kubernetes Operate section for detailed instructions.

Next Steps

• Instrument Your Application
• Use OpenTelemetry Operator for automatic instrumentation (if your applications are in k8s)
• Tutorials