Skip to content

Latest commit

 

History

History
26 lines (17 loc) · 2.28 KB

odfe-data-prepper.release-notes-0.8.0.md

File metadata and controls

26 lines (17 loc) · 2.28 KB

2021-03-25 Version 0.8.0 (beta)

Highlights:

  • Horizontally-scaling clusters can now be deployed by using the new Peer Forwarder plugin.
  • Prometheus-friendly metrics can now be scraped via a new /metrics endpoint.

Peer Forwarder Plugin

To enable horizontal scaling, it is expected to run Data Prepper instances in a cluster with requests distributed by a load balancer. This complicates preppers which maintain state between requests, specifically those which require that only one instance process all spans for a given trace. Peer forwarding facilitates this usecase by ensuring all spans for a given trace arrive at a specific Data Prepper instance.

The Peer Forwarding plugin enables instances in a Data Prepper cluster to forward requests to other specific instances within the cluster. Peer forwarding is accomplished by a consistent hashing algorithm to ensure that all requests for a given trace are routed to the same Data Prepper instance. The algorithm maps requests to addresses of known instances - these instances are provided via a peer discovery mechanism. As of version 0.8.0, peer discovery can be provided by either a static address list or by dynamic DNS lookups.

Refer to the README file for usage instructions.


Metrics

Version 0.8.0-beta offers APIs which provide JVM/system metrics and Plugin metrics.

Users can use these endpoints to connect to their existing Prometheus instance to start aggregating metrics. Additional tools such as Grafana can then be used to provide dashboards and alerting.

See the monitoring docs for a list of default metrics emitted by each component type (buffers, sinks, etc.)