Skip to content

Observability (SkyWalking)

Goals: end-to-end tracing; correlate Gateway → downstream spans; inject traceId into logs to enable trace ↔ log cross-navigation.
Instruments: Service / Endpoint metrics (QPS, Latency, Error), topology, trace details, dashboards.

1. End-to-End Overview

  • Service Overview: throughput, success rate, latency, Apdex across services.
  • Endpoint View: Top-N endpoints by traffic/success rate within the selected service to locate slow/error-prone APIs.

2. Service Topology

  • Visualize Gateway → services → DB/Cache/Vendor call relationships.
  • Solid vs. dashed lines distinguish real calls vs. telemetry reporting; supports depth control and failure-path filtering.

3. Trace Details: Async Task Path

  • Show the full span chain from enqueue → consume → external OpenAPI → persistence.
  • Common actions: jump from logs back to traces via traceId; inspect Tags/Logs/Stack on individual spans.

4. Service Dashboards: Key Metrics

  • Key metrics: avg/percentile latency (p50/p95/p99), QPS/error rate, per-instance load.
  • Aids: slow-endpoint ranking, MQ consumption delay, instance success rate, with selectable time windows.