Logging & OpenSearch Deployment

Introduction

Logs from different components present on the cluster will be pulled into OpenSearch to display dashboards and compute reports. Fluentd is used to pull captured logs and put into OpenSearch.

Prerequisites

• The following utilities/tools must be present on the user's machine.

  • kubectl, istioctl, helm, jq, curl, wget, git, bash, envsubst.

• Keycloak for Authentication and Sign-in to UI

Installation

Clone the https://github.com/openg2p/openg2p-deployment repo and navigate to kubernetes/logging directory.

Install OpenSearch (and related components)

• Run this to install OpenSearch and related components.

  SANDBOX_HOSTNAME="openg2p.sandbox.net" \
      ./install.sh

• After installation is successful, OpenSearch dashboards will be accessible at https://opensearch.openg2p.sandbox.net, depending on the hostname given above.

Install Rancher Logging (Fluentd)

1. On Rancher UI, navigate to Apps (or Apps & Marketplace) -> Charts

2. Search and install Logging from the list, with default values.

Add Index State Policy on OpenSearch

• Run this to add ISM Policy (This is responsible for automatically deleting logstash indices after 3 days. Configure the minimum age to delete indices, in the same script below.)

  ./opensearch-ism-script.sh

Configure Rancher Fluentd

• Run this to create ClusterOutput (This is responsible for redirecting all logs to OpenSearch.)

  kubectl apply -f clusteroutput-opensearch.yaml

• Run this to create a ClusterFlow (This is responsible for filtering OpenG2P service logs, from the logs of all pods.)

  kubectl apply -f clusterflow-all.yaml

Filters

Note the filters applied in clusterflow-all.yaml. You may update the same for your install if required, and rerun the apply command.

Dashboards

• TODO

TraceId

• TODO

Troubleshooting

• TODO