imbajin opened a new pull request, #838: URL: https://github.com/apache/hugegraph-toolchain/pull/838
PD can report `Cluster_OK` while partitions have `PState_Fault`, and Server `/metrics/backend` can return HTTP 200 with per-graph exceptions. Hubble previously treated both as healthy collection. This change shows PD cluster/data health separately from process status and marks backend metric failures with safe counts and actionable bilingual explanations. Cached metric values retain current failure counts; failed PD refreshes retain the last observed raw states with a stale marker. It also prevents configured-PD Prometheus metrics being attributed to an arbitrary leader when node identity cannot be matched after canonical scheme/host/default-port/path comparison, and prefers `/v1/prom/targets/store`, falling back to `targets-all` only for 404/405. Scheme-less PD member addresses use the configured scheme for matching; explicit schemes remain strict, aliases are not guessed, and existing node IDs are preserved. The identity-contract fixture covers the new route and 404/405 fallback. Existing target allowlists, credentials, response limits and request budgets remain unchanged. Validation: the complete Hubble UnitTestSuite passed 780 tests after synchronizing the diagnostics fixes; the final diagnostic-contract increment passed 93 backend and 37 frontend tests. On pinned Node 24.22.0, the three unrelated suites that exceeded the default 5-second local budget passed alongside all Operations and punctuation tests with a 15-second CLI budget (220 tests / 12 suites); no test files or CI timeout settings were changed. The local full frontend run was interrupted and is not claimed as passing. TypeScript, full ESLint, i18n, production build and 114 frontend/fixture script tests passed. Chrome authenticated/anonymous API fixtures verified PD data faults, failed-graph counts, stale states, diagnostics and context/account behavior; README screenshots identify the fixture limitation. Live cluster integration remains unverified, and current-head CI is pending. Depends on #82 and targets its branch so reviewers see only this health-contract change. Review/merge #82 first; then retarget this PR to master. Tracks Issues 9–11 and the Store-specific discovery portion of Issue 12 in the [issue record](https://chatgpt.com/space/page_62cf6a23a284819196000f124fdd6990). Readiness probes, lightweight Store topology and other documented lifecycle gaps remain follow-ups. --- ## Upstream submission notes Source community PR: https://github.com/hugegraph/hugegraph-toolchain/pull/83. This upstream PR reuses the same source repository and branch. Validation statements above are reports from the source task; they were not independently rerun during upstream submission. At submission preflight, source-head hosted checks: 15 skipped, 5 success. Upstream exact-head CI and current review feedback require separate confirmation; opening this PR does not establish merge or release readiness. Dependency: https://github.com/apache/hugegraph-toolchain/pull/837. The description above preserves the community PR context, including its original PR numbers and branch-base statement. This ASF PR instead targets master because the parent branch is not present upstream. Its current full master diff includes the pending parent changes (24 commits / 41 files); the incremental health-contract change against the parent is 12 commits / 17 files. Review and merge the parent first; this is not an independent master-only change. No upstream base branch was created. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
