Configuring Endpoints

Describes the endpoints in HPE Ezmeral Unified Analytics Software and how to configure them.

Configure endpoints in HPE Ezmeral Unified Analytics Software by going to Administration > Settings and selecting the Configurations tab.

The following sections provide details for each type of endpoint on the Configurations tab:

OTel Endpoint

The OTel endpoint is the target URL where HPE Ezmeral Unified Analytics Software OTel exporter sends metrics. The OTel endpoint enables other OTel collectors to receive cluster metrics in OTel format.

When you register an OTel endpoint, the cluster OTel collector exports metric data to the customer OTel collector hosted at the OTel endpoint. This includes Prometheus metrics about cluster performance, billing/metering related data, and app-based metrics for Kubeflow, Spark, and Ray. You can also export the incoming data to tools, such as Grafana or Elasticsearch.

OTEL is the standard format for metrics collection. Data only persists for 60 days in prometheus.

Use the following OTel endpoint format:
<host>:<port>
The OTel endpoint format:
  • Must be a valid HTTPS host
  • May contain a port
  • Should contain a path
  • Cannot contain other parts, such as a query string or fragment

JDBC Endpoint

The JDBC endpoint is automatically created when you install and configure HPE Ezmeral Unified Analytics Software.

To connect EzPresto to external applications, see Connecting External Applications to EzPresto via JDBC.

EzCentral Forwarding

NOTE
EzCentral Forwarding is not supported for air-gapped (disconnected) environments.

In HPE Ezmeral Unified Analytics Software, you can opt-in to forward your metrics collected by Prometheus to EzCentral via OTEL in real-time. EzCentral is a platform managed by HPE that can monitor your HPE Ezmeral Unified Analytics Software clusters when you enable EzCentral Forwarding.

The metrics forwarding to EzCentral has the following benefits:

  • Fast resolution of cluster issues by HPE through efficient cluster management and administration.
  • Provides real-time alerts to HPE enabling immediate incident resolution.
  • Enables HPE to warn you regarding unhealthy clusters and take actions to resolve issues.
  • Provides valuable metrics to fix bugs and improve user experience.

The metrics forwarding to EzCentral is enabled by default.

The forwarded metrics include:

  • Node – CPU or Memory or Disk Usage
  • Container – CPU or Memory
  • GPU usage
To disable metrics forwarding to EzCentral, toggle off the Metrics Opt In button as follows: