TMS Metrics

Deployment Guide (1.3.0)

TMS provides a metrics endpoint from which Prometheus formatted runtime metrics can be retrieved.

The following Helm chart options can be used to affect how metrics are reported:

  • server.metrics.enabled can be used to enable/disable metrics endpoint.

  • server.metrics.reportingWindow can be used to configure the reporting window for metrics endpoint.

  • server.metrics.minimumVisibility can be used to configure the which metrics are collected and reported.

By default, only high visibility are reported when metrics reporting is enabled.

Standard (high visibility) metrics are reported for each of the server’s endpoints.

Metric Name


tms_error_count Number of errored requests during the reporting window.
tms_duration_avg_seconds Average duration of successful requests during the reporting window.
tms_grpc_request_count Number of gRPC requests made during the reporting window.

Additional metrics are available by adjusting the minimum-visibility Helm chart value. These metrics are self-describing as part of the Prometheus formatted output.

© Copyright 2023, NVIDIA. Last updated on Dec 19, 2023.