Deviser Telemetry

This document describes the OpenTelemetry metrics and assets for Deviser.

For general setup instructions, see the OpenTelemetry Configuration guide.

Telemetry Assets

• dashboards/deviser-overview.dashboard.json: Grafana dashboard with panels for partition management, purge throughput, and storage monitoring.
• collector/otel-collector-config.yaml: Minimal OpenTelemetry Collector configuration.

Metrics Reference

Naming note

In Prometheus, metric names are normalized with underscores (e.g., deviser_startup_success), and counters typically include the _total suffix.

Metrics Table

Metric	Type	Unit	Description	PromQL example
deviser.startup.success	Counter	count	Service startup completed successfully.	sum(increase(deviser_startup_success_total[1h]))
deviser.startup.failure	Counter	count	Service startup failed.	sum(increase(deviser_startup_failure_total[1h]))
deviser.db.connect.duration_seconds	Histogram	seconds	Time spent waiting for PostgreSQL readiness.	histogram_quantile(0.95, sum by (le) (rate(deviser_db_connect_duration_seconds_bucket[5m])))
deviser.db.schema_apply.duration_seconds	Histogram	seconds	Time spent applying startup SQL schemas.	histogram_quantile(0.95, sum by (le) (rate(deviser_db_schema_apply_duration_seconds_bucket[5m])))
deviser.db.schema_apply.failure	Counter	count	Failures during schema application.	sum(increase(deviser_db_schema_apply_failure_total[1h]))
deviser.partitions.ensure.duration_seconds	Histogram	seconds	Duration of daily partition creation routine.	histogram_quantile(0.95, sum by (le) (rate(deviser_partitions_ensure_duration_seconds_bucket[5m])))
deviser.partitions.ensure.failure	Counter	count	Failures in partition ensure routine.	sum(increase(deviser_partitions_ensure_failure_total[1h]))
deviser.partitions.ensure.days	Gauge	days	Configured horizon for partition creation.	max(deviser_partitions_ensure_days)
deviser.partitions.maintain.duration_seconds	Histogram	seconds	Duration of partition maintenance routine.	histogram_quantile(0.95, sum by (le) (rate(deviser_partitions_maintain_duration_seconds_bucket[5m])))
deviser.partitions.maintain.failure	Counter	count	Failures in partition maintenance.	sum(increase(deviser_partitions_maintain_failure_total[1h]))
deviser.partitions.dropped.expired	Counter	count	Partitions dropped due to retention policy.	sum(increase(deviser_partitions_dropped_expired_total[1h]))
deviser.partitions.dropped.quota	Counter	count	Partitions dropped due to quota enforcement.	sum(increase(deviser_partitions_dropped_quota_total[1h]))
deviser.partitions.bytes_freed	Counter	bytes	Bytes reclaimed by dropping partitions.	sum(increase(deviser_partitions_bytes_freed_total[1h]))
deviser.partitions.total_discovered	Gauge	count	Number of partitions discovered during maintenance.	max(deviser_partitions_total_discovered)
deviser.partitions.total_bytes	Gauge	bytes	Total size in bytes of discovered partitions.	max(deviser_partitions_total_bytes)
deviser.resources.purge.duration_seconds	Histogram	seconds	Duration of soft-delete purge routine.	histogram_quantile(0.95, sum by (le) (rate(deviser_resources_purge_duration_seconds_bucket[5m])))
deviser.resources.purge.rows	Counter	rows	Number of rows hard-deleted from krateo_resources.	sum(increase(deviser_resources_purge_rows_total[1h]))
deviser.resources.purge.failure	Counter	count	Failures in soft-delete purge routine.	sum(increase(deviser_resources_purge_failure_total[1h]))
deviser.loop.iteration.success	Counter	count	Main scheduled loop iteration with no errors.	sum(rate(deviser_loop_iteration_success_total[5m]))
deviser.loop.iteration.failure	Counter	count	Main scheduled loop iteration with at least one error.	sum(rate(deviser_loop_iteration_failure_total[5m]))