JiaLiangC opened a new pull request, #4216: URL: https://github.com/apache/ambari/pull/4216
Issue: [AMBARI-26656](https://issues.apache.org/jira/browse/AMBARI-26656) ## What changes were proposed in this pull request? Allow one Ambari console to manage independently owned clusters while retaining the current cluster's Dashboard. Admin gains a cluster overview, creation, host allocation, scoped permissions, and versions/repositories navigation. Each cluster keeps its own hosts, services, requests, browser state and realtime event access. The backend owns recoverable HBase dependency deployment state: persistent binding/operation/epoch and request/task lineage connect REST, ActionManager, heartbeat results and credential completion. Blueprint and Add Service share managed dependency constraints, and Agent helpers use typed Hadoop/ZooKeeper/HBase SDK observations. Native testing also fixes Registry DNS supervisor shutdown, command-scoped HBase hook initialization, nested JSON routes and Admin View session startup. This submission consolidates the work originally reviewed under AMBARI-26654 under the new issue requested for publication. It is based on current trunk after #4208, with 27 topic commits plus the publication record. The initial foundation keeps new server/ORM/API contracts with their required frontend and agent consumers; subsequent isolation, recovery, UI, runtime and evidence corrections are separate reviewable commits. The original review branch and packaged-source SHAs are retained. See [the source review, original findings, remediation ledger and exact validation commands](https://github.com/JiaLiangC/ambari/blob/AMBARI-26656/docs/design/multi-cluster-review-and-remediation.md). ## How was this patch tested? **On the submission branch after rebasing onto trunk:** 195 YARN and 22 Hive Python tests passed, and `git diff --check apache/trunk..HEAD` passed. The source-tree delta from the previously accepted review HEAD is exactly the seven upstream baseline corrections (verified by stable patch ID), plus this documentation record. ```shell PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=ambari-common/src/main/python:ambari-server/src/main/resources/stacks python3 -B -m unittest discover -s ambari-server/src/test/python -p 'TestYarnBigtop*.py' -q PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=ambari-common/src/main/python:ambari-server/src/main/resources/stacks python3 -B -m unittest discover -s ambari-server/src/test/python -p 'TestHiveChecksBigtop.py' -q ``` **Before the publication-only rebase:** 120 focused Java tests, 116 HBase/managed-dependency Python tests, 42 unique frontend tests and both frontend builds passed. The subsequent Registry DNS/HBase-hook corrections passed 192 YARN tests and 2 hook tests, plus live recovery checks. Exact commands and checkpoint-specific limitations are in the linked review record; these counts are not claims that the full suite was rerun at the PR SHA. **Native deployment:** deploy built RPMs from source `1ab17a10ea333e35d9952666d57d8f4383762103`, release `1789036336.git1ab17a10ea33`. Six Agents registered; A used three nodes and the remaining three were initially unassigned. API creation of B completed, then ordinary API Add Service installed HBase in both clusters. Requests 19/23 and SDK read/write checks 22/26 completed. Effective roots were `hdfs://mc-local-a/apps/hbase/data` and `hdfs://mc-local-b/apps/hbase/data`, with distinct HBase UUIDs and each cluster's own ZooKeeper hosts. All 44 daemon components were STARTED and 26 clients INSTALLED. Four foreign GET/PUT operations returned 403, duplicate host attachment returned 409, and a foreign scoped request returned 404. Seven installed RPM identities and 21 runtime script hashes matched the packaged source; no overlays were applied to that final deployment. Before final packaging, two B stop/start cycles, A HBase read/write while B was stopped, and Ambari Server restart persistence passed after partial corrections. Earlier managed-topology tests also completed cross-cluster HBase SDK read/write and real scoped STOMP/browser checks. Final RPM screenshots below show presentation only; API/SDK receipts establish behavior. **Remaining limits:** no real KDC callback acceptance, complete managed provider stop/delete/detach retention matrix, all crash/response-loss boundaries, complete broker revocation matrix, or full test-suite/all-database validation. Monitoring has no configured Prometheus target. No fresh RPM build/deployment is claimed for the seven upstream corrections in the rebased submission branch. **Two installed clusters, each with HBase and its own Hadoop services**  **Six registered hosts with exclusive three-host cluster ownership**  Please review [Ambari Contributing Guide](https://cwiki.apache.org/confluence/display/AMBARI/How+to+Contribute) before opening a pull request. -- 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]
