O11: Grafana dashboard for memory-observability #37

Open
opened 2026-09-08 05:02:17 +00:00 by rock · 0 comments
Owner

Build Grafana dashboard with 10 rows covering all observability signals.

Rows:

  1. System Health — health status, PG pool, OpenSearch, embedding model
  2. Write Path (Ingest) — rate, latency, dedup ratio, OS failures, contradiction queue
  3. Read Path (Query) — rate, latency breakdown (sem/lex/rrf/rerank), empty results, score dist
  4. Context (3-Tier) — tier hit distribution, latency, budget usage, degraded responses
  5. Relevance (Qwen-7B) — NDCG@10, MRR, Precision/Recall, judge cost, agreement
  6. Pod Resources — memory usage + limit, CPU + throttle, restarts, network I/O
  7. Availability — availability ratio, error rate, dependency status, degraded responses
  8. Ingest Rate Patterns — write rate/min + /hour, by project/level, queue depth/age
  9. Postgres Internals — cache hit ratio, dead tuples, connections, seq vs idx, replication lag, WAL, slowest queries
  10. Storage — total vectors, bytes, parity drift, level distribution, compaction freed

Export dashboard JSON to k8s/monitoring/grafana-dashboard-memory.json

Reference: docs/OBSERVABILITY.md Section 11

Build Grafana dashboard with 10 rows covering all observability signals. Rows: 1. System Health — health status, PG pool, OpenSearch, embedding model 2. Write Path (Ingest) — rate, latency, dedup ratio, OS failures, contradiction queue 3. Read Path (Query) — rate, latency breakdown (sem/lex/rrf/rerank), empty results, score dist 4. Context (3-Tier) — tier hit distribution, latency, budget usage, degraded responses 5. Relevance (Qwen-7B) — NDCG@10, MRR, Precision/Recall, judge cost, agreement 6. Pod Resources — memory usage + limit, CPU + throttle, restarts, network I/O 7. Availability — availability ratio, error rate, dependency status, degraded responses 8. Ingest Rate Patterns — write rate/min + /hour, by project/level, queue depth/age 9. Postgres Internals — cache hit ratio, dead tuples, connections, seq vs idx, replication lag, WAL, slowest queries 10. Storage — total vectors, bytes, parity drift, level distribution, compaction freed Export dashboard JSON to k8s/monitoring/grafana-dashboard-memory.json Reference: docs/OBSERVABILITY.md Section 11
rock added the memory-observability label 2026-09-08 05:02:17 +00:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: riotpiao-poimen/poimen-memory#37