Self-Observability Metrics¶

SemConv Proxy exposes 30+ Prometheus metrics at /metrics for monitoring the proxy itself.

Endpoint¶

GET /metrics

Returns Prometheus text exposition format on the API port (default: 8080).

Metric Categories¶

Signal Throughput¶

Metric	Type	Labels	Description
`semconv_proxy_signals_received_total`	counter	`signal_type`, `protocol`	Total OTLP signals received
`semconv_proxy_signals_forwarded_total`	counter	`signal_type`, `protocol`	Total signals forwarded to backend
`semconv_proxy_signals_dropped_total`	counter	`signal_type`, `protocol`	Total signals dropped (backend failure)

Label values:

signal_type: metric, trace, log
protocol: http, grpc

Example query — received vs forwarded:

rate(semconv_proxy_signals_received_total[5m])
-
rate(semconv_proxy_signals_dropped_total[5m])

Dictionary Operations¶

Metric	Type	Labels	Description
`semconv_proxy_dictionary_entries`	gauge	—	Current number of dictionary entries
`semconv_proxy_dictionary_attributes_added_total`	counter	—	Total attributes added
`semconv_proxy_dictionary_attributes_changed_total`	counter	—	Total attributes changed
`semconv_proxy_dictionary_attributes_removed_total`	counter	—	Total attributes removed

Example query — attribute growth rate:

rate(semconv_proxy_dictionary_attributes_added_total[1h])

Cardinality¶

Metric	Type	Labels	Description
`semconv_proxy_cardinality_high_attributes`	gauge	—	Number of high-cardinality attributes
`semconv_proxy_cardinality_budget_utilization`	gauge	—	Global budget utilization percentage

Example alert — budget approaching limit:

semconv_proxy_cardinality_budget_utilization > 80

Analysis Pipeline¶

Metric	Type	Labels	Description
`semconv_proxy_pipeline_ring_buffer_size`	gauge	—	Current ring buffer fill level
`semconv_proxy_pipeline_lag`	gauge	—	Analysis pipeline lag
`semconv_proxy_pipeline_drops_total`	counter	—	Total dropped analysis tasks
`semconv_proxy_pipeline_processing_duration_seconds`	histogram	`stage`	Processing time per stage

Example query — p99 processing time:

histogram_quantile(0.99, rate(semconv_proxy_pipeline_processing_duration_seconds_bucket[5m]))

Storage¶

Metric	Type	Labels	Description
`semconv_proxy_storage_persist_duration_seconds`	histogram	`operation`	Pebble persist duration
`semconv_proxy_storage_disk_size_bytes`	gauge	—	Pebble disk size in bytes

API¶

Metric	Type	Labels	Description
`semconv_proxy_api_request_total`	counter	`path`, `method`, `status`	Total API requests
`semconv_proxy_api_request_duration_seconds`	histogram	`path`	API request duration

Example query — API error rate:

rate(semconv_proxy_api_request_total{status=~"5.."}[5m])
/
rate(semconv_proxy_api_request_total[5m])

Prometheus Scrape Config¶

scrape_configs:
  - job_name: semconv-proxy
    static_configs:
      - targets:
          - semconv-proxy.observability:8080
    metrics_path: /metrics
    scrape_interval: 15s

Grafana Dashboard¶

A minimal dashboard might include these panels:

Panel	Query	Visualization
Signal Throughput	`rate(semconv_proxy_signals_received_total[5m])`	Time series
Forwarding Success Rate	`rate(semconv_proxy_signals_forwarded_total[5m]) / rate(semconv_proxy_signals_received_total[5m])`	Stat
Dictionary Size	`semconv_proxy_dictionary_entries`	Stat
Cardinality Budget	`semconv_proxy_cardinality_budget_utilization`	Gauge
Pipeline Drops	`rate(semconv_proxy_pipeline_drops_total[5m])`	Time series
API Latency p99	`histogram_quantile(0.99, rate(semconv_proxy_api_request_duration_seconds_bucket[5m]))`	Time series