Troubleshooting Data Streams Monitoring

This page is not yet available in Spanish. We are working on its translation.
If you have any questions or feedback about our current translation project, feel free to reach out to us!

Data Streams Monitoring is not available for the site.

This page explains common issues with setting up and using Data Streams Monitoring, and how to resolve them. Datadog recommends staying up to date with the latest version of the Datadog tracing libraries you use, as each release contains improvements and fixes.

Diagnosing common problems

Services aren’t showing up in the DSM Map

If you do not see your services in the DSM Map or Overview page after following the setup instructions, ensure these requirements are met:

  • You’re running the Datadog Agent v7.34.0 or later.
  • Your service is directly producing or consuming from Kafka or RabbitMQ.
  • You’re running the following tracing library agent versions:
    • Java: Agent v1.9.0 or later
    • .NET: Tracer v2.28.0 or later (.NET Core, .NET Framework)
    • Go (manual instrumentation): Data Streams Library v0.2 or later

End-to-end latency metric doesn’t look accurate

Latency calculations on a pathway require messages traversing through the pathway to be single threaded. If the messages in your data pipelines are multi-threaded, manual instrumentation is required, which is currently available for Go applications and Java applications. If you require manual instrumentation for .NET, contact Support.

In the Pathways tab, if you see the message latency values may be approximate for these pathways, view the documentation for instrumentation guidelines.

Kafka metrics are missing

If your application is running in Java, ensure you’re running version v1.9.0 or later of the Java Agent, and that both the producer and the consumer services are instrumented.

If your applications are running in .NET or Go, the Kafka Consumer Integration is required for the Kafka metrics to populate.

RabbitMQ metrics are missing

Ensure the RabbitMQ integration is set up correctly.

Queue metrics are missing

The Kafka integration must be set up for self-hosted, MSK, and Confluent Platform/Cloud environments for metrics to populate on the Queue tab.

Cluster tag is not appearing

The cluster tag is set differently depending on the environment:

  • Self-hosted Kafka: The kafka_cluster tag must be added to the configuration of your agent running in the same cluster as your Kafka brokers, with the key set to kafka_cluster and value set to your cluster name.
  • Amazon MSK: Cluster information is automatically propagated to DSM if the MSK Integration is turned on. MSK sends the cluster to DSM as cluster_name.
  • Confluent Cloud: Cluster information is automatically propagated to DSM if the Confluent Cloud Integration is set up on the cluster you instrumented with DSM.
  • Confluent Platform: Similar to self-hosted Kafka above, the kafka_cluster tag must be added to your agent configuration.
PREVIEWING: rtrieu/product-analytics-ui-changes