Configure Log Collection, Target, and Visualization - GCP

Note!

You need to have a proper GKE cluster setup in order to proceed with these steps. Refer to Set Up Kubernetes Cluster - GCP (4.3) to create the GKE cluster first.

By default Usage Engine deployed in Kubernetes outputs logging to disk and console output. If persistent disk storage is enabled, the logs end up on the mounted shared disk. But persistent disk is not always the desired log target, especially in a cloud environment where persistent data is typically accessed through services and APIs rather than as files. The console logs can be accessed through the "kubectl logs" command or from a Kubernetes dashboard. The buffer for storing the Kubernetes console logs is stored in memory only though and thus will be lost when a Pod terminates.

To get a production ready log configuration you can use tools from the Kubernetes ecosystem and GCP Cloud Logging. In this guide we show you how to set up:

GCP Cloud logging for storage and monitoring
Fluent-bit for log collection and log forwarding
Elasticsearch for log storage
Kibana for log visualization

These tools give you powerful and flexible log collection, storage, monitoring and visualization. The Elasticsearch database storage also provides powerful tools to perform analytics on the log data. The GCP Logs Explorer is a monitoring service built for DevOps engineers, developers, site reliability engineers (SREs), IT managers, and product owners. This guide doesn't describe these tools' functionality in details as it is outside the scope of this guide.

Prerequisite

Before setting up log collection, make sure your Usage Engine Private Edition was installed with JSON formatted logging enabled.

log:
  # Format can be "json" or "raw". Default is "raw"
  format: json

Stream container logs to GCP Cloud Logging

Before using GCP Cloud Logging, you need to ensure Cloud Logging API is enabled on your Google Cloud project. Refer to the guide https://cloud.google.com/kubernetes-engine/docs/troubleshooting/logging to verify if logging is enabled.

Fluent-bit is a log processor that used to send containers logs to GCP Cloud Logging. By default, a managed Fluent-bit will be installed by GKE during cluster creation.

After Cloud Logging API is enabled, all containers logs should automatically send to the Cloud Logging. To verify logging, go to GCP console page Logging > Logs Explorer and check if container logs are populated.

Stream container logs to Elastic Search and visualize with Kibana

Note!

Note that you must install Elastic Search, Fluent-bit and Kibana on the same namespace in order to allow working properly. There are some of the reasons:

Elastic Search service needs to be accessible by Fluent-bit and Kibana to establish connection.
Kibana required Elastic Search master cert secret presented on the namespace.

Hence, in this guide we are using namespace 'logging' for the installations.

Install Elastic Search

Add Elastic Search repository to Helm and update repository to retrieve the latest version.
```
helm repo add elastic https://helm.elastic.co
helm repo update
```
Install Elastic Search.
Note!

For simplicity this example installs Elasticsearch without persistent storage. Refer to Elasticsearch Helm chart documentation for help to enable persistent storage:

https://github.com/elastic/helm-charts/tree/master/elasticsearch
```
helm install elasticsearch elastic/elasticsearch -n logging --set=persistence.enabled=false
```

Install custom Fluent-bit

Add Fluent helm repository and update repository to retrieve the latest version.
```
helm repo add fluent https://fluent.github.io/helm-charts
helm repo update
```

Retrieve the Elastic Search access credentials by using commands below. Save the output, you will need them in the next step.

kubectl get secrets --namespace=logging elasticsearch-master-credentials -ojsonpath='{.data.username}' | base64 -d
kubectl get secrets --namespace=logging elasticsearch-master-credentials -ojsonpath='{.data.password}' | base64 -d

Create a custom values yaml, for example fluent-bit-values.yaml and set the content below. After that, replace values of HTTP_User and HTTP_Passwd to the output from previous step.

config:
  inputs: |
    [INPUT]
        Name                tail
        Alias               kube_containers
        Tag                 kube_<namespace_name>_<pod_name>_<container_name>
        Exclude_Path        /var/log/containers/*_kube-system_*.log,/var/log/containers/*_istio-system_*.log,/var/log/containers/*_knative-serving_*.log,/var/log/containers/*_gke-system_*.log,/var/log/containers/*_config-management-system_*.log,/var/log/containers/*_gmp-system_*.log,/var/log/containers/*_gke-managed-cim_*.log
        Path                /var/log/containers/*.log
        multiline.parser    docker, cri
        Mem_Buf_Limit       50MB
        Skip_Long_Lines     On
        Refresh_Interval    1
        Read_from_Head      True

  filters: |
    [FILTER]
        Name                kubernetes
        Match               kube.*
        Kube_URL            https://kubernetes.default.svc:443
        Kube_Tag_Prefix     application.var.log.containers.
        Merge_Log           On
        K8S-Logging.Parser  On
        K8S-Logging.Exclude Off
        Labels              Off
        Annotations         Off
        Use_Kubelet         On
        Kubelet_Port        10250
        Buffer_Size         0

  outputs: |
    [OUTPUT]
        Name                es
        Match               *
        Host                elasticsearch-master
        tls                 on
        tls.verify          off
        HTTP_User           elastic
        HTTP_Passwd         zUqEBtrE4H9bfO8K
        Suppress_Type_Name  On
        Index               fluentbit
        Trace_Error         on

Install Fluent-bit with the custom values yaml.

helm install fluent-bit fluent/fluent-bit -n logging -f fluent-bit-values.yaml

Verify Fluent-bit pod's log. Should not see any error or exception if connection to Elastic Search is established successfully.
```
kubectl logs <fluent-bit pod name> -n logging
```

Install Kibana

Install Kibana. Note that service type is set to LoadBalancer in order to allow public access.

helm install kibana elastic/kibana -n logging --set=service.type=LoadBalancer --set=service.port=80

Configure Kibana

Kibana is a visual interface tool that allows you to explore, visualize, and build a dashboard over the log data massed in Elastic Search cluster.

Up to this stage, all pods under namespace logging should be up and running.

NAME                             READY   STATUS    RESTARTS   AGE
elasticsearch-master-0           1/1     Running   0          4d3h
elasticsearch-master-1           1/1     Running   0          4d3h
elasticsearch-master-2           1/1     Running   0          4d3h
fluent-bit-2kpgr                 1/1     Running   0          3d
fluent-bit-6wtnr                 1/1     Running   0          3d
fluent-bit-ns42z                 1/1     Running   0          3d
kibana-kibana-658dc749cd-hbc8s   1/1     Running   0          3d4h

If all looks good, you can proceed to login to Kibana dashboard web UI.

Retrieve the public access IP of the Kibana dashboard.

kubectl get service -n logging kibana-kibana -o jsonpath='{.status.loadBalancer.ingress[0].ip}'

Login to Kibana dashboard web UI with username password same as HTTP_User and HTTP_Passwd configured in previous section.
Go to Management > Stack Management > Index Management. Create the Index Template with Index Pattern matching the indexes configured in previous section.
If Fluent-bit connection to Elastic Search established successfully, the Indices is created automatically.
Go to Management > Stack Management > Kibana. Create Data view matching the index pattern
Go to Analytics > Discover to view logs.
User can filter logs using KQL syntax. For instance, enter "ECDeployment" in the KQL filter input field.