Self-telemetry: every component ingests to its own HoldFast instance

[BrewingCoder]

Summary

HoldFast should eat its own dogfood. Every component in a HoldFast deployment — backend, frontend, and infrastructure — should automatically send telemetry to the HoldFast instance it's running alongside. No manual configuration required.

What's already done (this issue tracks the remainder)

• SystemBootstrapService — creates the HoldFast workspace + project on every startup (idempotent, always runs)
• .NET backend OTeL SDK wired — traces + metrics via OTLP to OTEL_EXPORTER_OTLP_ENDPOINT (defaults to http://localhost:8082)
• Docker Compose: OTEL_EXPORTER_OTLP_ENDPOINT=http://backend:8082 set for backend container
• Login response shape fixed ({token, user: {uid, email}})

Remaining work

Frontend browser SDK

• Add @holdfast-io/browser to src/frontend/
• Initialize in src/frontend/src/index.tsx using REACT_APP_PUBLIC_GRAPH_URI and REACT_APP_HIGHLIGHT_PROJECT_ID (injected at runtime)
• This requires issue Frontend: replace baked-in URL hack with runtime config injection #63 (runtime config injection) to be resolved first

Backend log export

• Add OpenTelemetry.Exporter.OpenTelemetryProtocol log exporter alongside traces/metrics
• Wire ILogger → OTeL → /otel/v1/logs

Infrastructure exporters (Docker Compose)

• PostgreSQL: add postgres-exporter sidecar → Prometheus → OTeL collector → HoldFast
• ClickHouse: add clickhouse-exporter → HoldFast
• Kafka: add JMX exporter or kafka-exporter → HoldFast
• Redis: add redis-exporter → HoldFast

OTeL Collector (optional)

• Add otel-collector container as a central aggregation point for infra exporters
• Routes all infra metrics to the HoldFast public endpoint

Helm / k8s

• Mirror Docker Compose telemetry wiring in Helm chart values (tracked in Helm charts + MicroK8s deployment path #62)

Design principle

When someone installs HoldFast on-prem, they should open the dashboard after first boot and already see their own platform's health data. Zero additional configuration.

[github-actions]

AI Triage: Implement self-telemetry for all HoldFast components (frontend SDK, backend logs, infrastructure exporters) to send data to the instance's own HoldFast deployment with zero manual configuration.

Labels applied: enhancement, area:backend, priority:high, component:traces