Aias00 opened a new issue, #7720: URL: https://github.com/apache/shenyu/issues/7720
## Goal Make Admin an entry point to existing observability tools from instance diagnostics. Keep high-volume metrics, logs and spans in the configured backend rather than storing them in heartbeat payloads or the Admin relational database. ## Current boundary The instance page has presence/basic machine information. General tracing is tracked by #7566, and #7614 covers tracing plugin control-plane registration rather than querying trace data. Agent Gateway already has an optional [logical MCP call span bridge](https://github.com/apache/shenyu/blob/66d7d954ca37d651a5ad916f988a8081157a7ab4/shenyu-plugin/shenyu-plugin-agent-gateway/README.md#L512); enabling that bridge alone does not supply SDK export, inbound/outbound propagation or an Admin trace explorer. ## Initial scope - Define correlation metadata shared by status reports, metric resources, logs and traces: namespace, service/deployment, stable instance identity and boot identity where supported. - Add operator-configured metric/log/trace navigation links in instance detail with explicit time-window filters. For exact request links use an actual trace ID from a request/log record; do not invent one for an instance. - Support existing Grafana/Prometheus, logging and tracing deployments through configured URL templates/adapters. Begin with navigation, not a new analytics storage engine. - Display collection/SDK configuration only to the extent it can be observed reliably. Do not claim successful end-to-end telemetry delivery merely because an exporter is configured. ## Acceptance criteria - [ ] A test fixture correlates one gateway instance across its status, metrics, access log and exported trace using documented identifiers. - [ ] Selected namespace/instance/time range is correctly encoded in configured links; no credentials or sensitive payload data are embedded in URLs. - [ ] Missing backend configuration and unavailable data have clear states and do not break status pages or proxy traffic. - [ ] Namespace authorization is retained; navigation does not pretend to grant access to the destination platform. - [ ] Dashboard changes have tests for filtering, link generation and legacy instances without correlation fields. - [ ] Metric label cardinality is budgeted; raw trace IDs/request paths are not added as general metric labels. Related: #7566/#7614 and #7650. Reuse deployed instrumentation and backend integration rather than duplicating tracing implementation. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
