Repository navigation
Conversation
bitflicker64
left a comment
There was a problem hiding this comment.
Blocking: no. Score: 9/10. Summary: The PD bootstrap and quorum rules, drain retry, 30-second metadata LOCK wait, 1 GiB raft.maxSegmentFileSize default, blank-segment isolation, Store recovery NOT_FOUND handling and learner catch-up text match StoreNodeService, ConfigService, HgKVStoreImpl, RaftSegmentLogStorage, PartitionManager and PartitionEngine on Server master; the Server and Docker diagnostics rewrite keeps the launcher behaviour of hugegraph-server.sh.
imbajin
left a comment
There was a problem hiding this comment.
Score: 8.5/10 · Confidence: medium
Blocking: no
Summary
No new actionable issues found in the storage recovery and deployment guidance.
Evidence
Reviewed the bilingual changes and discussion, and checked recovery, segment defaults, membership changes and launcher guidance against Server source. Current documentation build and visual CI jobs passed.
Validation notes
No live Store recovery, migration, Docker or Kubernetes test was run during this review. Source inspection does not establish failure-recovery behavior in a running cluster.
Purpose of the PR
Give operators accurate storage and deployment guidance where old defaults or overly strong recovery statements could lead to the wrong action.
Validation: source-contract and independent cross-review passed, along with strict artifact/link checks and desktop/mobile table-reader/Details checks. Existing routes, headings and version distinctions are retained.
Recovery and quorum statements were checked against source; the earlier minimal Helm run does not establish live HA or failure-recovery acceptance.