You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Upgrade Notes
This release advances Doctor from v0.1.33 to v0.1.80 and includes a substantial evolution of the Plugin SDK and Collect protocol. Plugin authors should adopt service-owned capabilities and the typed Fact, Observation, and Finding contracts before requiring this version.
New Features
Unified service diagnosis around a single Inspect → Probe → Detector pipeline. Core and Plugin contributions now share typed evidence identities, schema versions, provenance, access declarations, and deterministic execution through runCollect.
Added workload-aware doctor inspect, including Deployment environment and ConfigMap evidence, runtime workload discovery, dependency inspection, service environment probes, and AppArmor admission checks.
Added tenant inventory and intention collection, plus identity-scoped capability Facts and relations for Plugin-defined tenant data.
Added case-driven doctor eval and a canonical CaseSet shared with Perf, allowing each case to trigger a real request and collect its correlated Trace, Log, and Data evidence.
Standardized report delivery across commands with structured formats, HTML and Evidence Bundle output, bundle manifests, one-root extraction, namespace guidance, and generated bundle-reading instructions.
Expanded store diagnostics with MinIO storage-full detection, S3 prefix object counts, service-owned store targets, and richer Redis, database, VDB, and S3 evidence.
Added structured output for model diagnosis, remembered model selections, Doctor host details, Kubernetes server version, paginated data Facts, and improved Fact tables.
Bug Fixes
Kubernetes logs are now streamed through the client API, with concurrent service collection and clearer access checks.