Agent v1.35.0

Multiprocessor nodes, enabling multiple processors within a single, Live Capture support to allow real-time data ingestion, new Google Cloud Logging output node that uses, and Released the demo template source to Beta.

  2 minute read  

April 18, 2025

New Features

  • Introduced Multiprocessor nodes, enabling multiple processors within a single node for advanced pipeline processing.
  • Added Live Capture support to allow real-time data ingestion and inspection within pipelines.
  • Added a new Google Cloud Logging output node that uses OTTL expressions for configuration; the previous node is now marked as Beta for existing users.
  • Released the demo template source to Beta.

Enhancements

  • Updated OTLP conversion logic to group items by resource hash (instead of service name), use hex decoding for IDs, include span events, and improve scope batching to support mixed scopes.
  • Made the json_field_path field required in the JSON Unroll node for consistency with node validation requirements.
  • Added validation to the HTTP Pull endpoint setting to prevent leading and trailing spaces that could cause failures.
  • Added failure path handling for additional processors, improving reliability of multiprocessor node construction.
  • Enabled reporting of AWS S3 and Azure Blob storage traffic via self-telemetry, addressing discrepancies in metrics dashboards and editors.
  • Improved error visibility by reporting processor and strategy errors with self-telemetry.
  • Made name and interval required fields in the aggregate metrics processor for more consistent configuration validation.
  • Improved OTTL support for nodes:
    • Updated the Prometheus exporter output node to use OTTL expressions for custom metric labeling, enabling flexible configurations.
    • Updated the Loki output node to use OTTL expressions for custom labels, allowing dynamic configuration.
    • Updated AWS Cloudwatch output node to support OTTL expressions for fetching data attributes, with backward compatibility for CEL.

Bug Fixes

  • Fixed missing ED output data in the cost optimization API by ensuring outputs are checked in both output and ED output node lists.
  • Prevented duplicate data in sequence processors by avoiding passthrough when termination or condition mismatch occurs.
  • Fixed processor data type filtering so that data is not dropped when the incoming type does not match the processor’s supported types.
  • Fixed double logging in some cases for compactor receive errors by improving error handling.
  • Tightened validation of custom tags in the Sumo Logic Pusher by dropping invalid key-value pairs before sending.
  • Fixed config fetch error handling so errors are now properly reported instead of being ignored.
  • Fixed processor data type handling so unsupported item types pass through the processor rather than being dropped.