HDDS-15941. Resolve client before block loop and return null from sortDatanodesForWrite

[chihsuan]

What changes were proposed?

Follow-up to #10633.

• Updates OM write-path datanode sorting to resolve the client once.
• Let SCM sort if OM cannot find the client's rack.
• Return null when OM skips sorting because a pipeline datanode is missing from its topology, preserving the pipeline's order.

JIRA

https://issues.apache.org/jira/browse/HDDS-15941

Testing

• TestOMAllocateBlockRequest and TestOMSortDatanodes passed

https://github.com/chihsuan/ozone/actions/runs/34728833801

Generated-by: Claude Code (Claude Opus 5)

[Copilot]

🟢 Approval recommended

No unresolved issues were identified that would block approval.

Pull request overview

Updates OM write-path datanode sorting to resolve the client once, delegate unresolved clients to SCM, and preserve pipeline order when topology data is incomplete.

Changes:

• Added client resolution before block allocation.
• Changed skipped write sorting to return null.
• Expanded unit and integration tests for fallback, caching, and topology cases.

File summaries

File	Description
hadoop-ozone/ozone-manager/src/test/java/org/apache/hadoop/ozone/om/request/key/TestOMAllocateBlockRequest.java	Updated as part of this pull request.
hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/request/key/OMKeyRequest.java	Updated as part of this pull request.
hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/KeyManagerImpl.java	Updated as part of this pull request.
hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/KeyManager.java	Updated as part of this pull request.
hadoop-ozone/integration-test/src/test/java/org/apache/hadoop/ozone/TestOMSortDatanodes.java	Updated as part of this pull request.

Review details

• Files reviewed: 5/5 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[rich7420]

Please use a separate metric for client resolution. ConcurrentMutableRate exports one NumOps/AvgTime pair, so recording the once-per-request lookup here and every uncached pipeline sort into AllocateBlockSortDatanodesLatencyNs makes the count and average represent neither operation. Please add AllocateBlockResolveClientLatencyNs and keep the existing metric for sorting.