Triton+TRT-LLM TRT-LLM inflight-batcher gauge, label-switched
triton-inference-server · gauge
Fields
id
metric-exposures/w3m-triton-inference-server-nv-trt-llm-inflight-batcher-metricsMetric name
nv_trt_llm_inflight_batcher_metrics
Metric type
gauge
Unit
1
Exposure
prometheus
Endpoint
:8002/metrics
Labels
modelversioninflight_batcher_specific_metric
Enabled by default
yes
notes
Triton main, docs/user_guide/metrics.md, read 2026-10-04. On by default: 'By default, these metrics are available at http://localhost:8002/metrics', and the endpoint is pure pull - 'not pushed or published to any remote server'. tritonserver --allow-metrics=false turns the whole thing off. Labels are model and version, and crucially the LATENCY families are per MODEL, per request-update - not per-request observations as in vLLM/SGLang/TGI. Backend-contributed via the Triton custom metrics API, NOT a Triton core metric. All six families share the model and version labels. The label values are not enumerated in the README sample output, so the exact key set is deployment-dependent; read the endpoint rather than assuming a fixed list.
Notes
Triton main, docs/user_guide/metrics.md, read 2026-10-04. On by default: 'By default, these metrics are available at http://localhost:8002/metrics', and the endpoint is pure pull - 'not pushed or published to any remote server'. tritonserver --allow-metrics=false turns the whole thing off. Labels are model and version, and crucially the LATENCY families are per MODEL, per request-update - not per-request observations as in vLLM/SGLang/TGI. Backend-contributed via the Triton custom metrics API, NOT a Triton core metric. All six families share the model and version labels. The label values are not enumerated in the README sample output, so the exact key set is deployment-dependent; read the endpoint rather than assuming a fixed list.