Cezar Moise

Unfortunately, Jenkins failed at log collection so full logs not available but `system.log` for nodes is available. \I'm running it again to try to get all logs.

> Only node1 has debug-level load balancer logs, but it's not a leader any more at the time of event, so I don't have visibility into load balancer state. It...

@tgrabiec Replicated [Argus](https://argus.scylladb.com/tests/scylla-cluster-tests/e20f60d1-8826-431d-abf9-4639541ef5a8) [Monitoring](http://3.251.71.149:3000/) Still I don't see debug logs in the nodes that were started later, but the nodes are started with the command line argument added ``` 2025-04-08T23:57:53.022+00:00...

> Even if we fix https://github.com/scylladb/scylladb/issues/22982, temporary imbalance is expected in this scenario, unless we block table creation on rebalancing. For example, recommend users to wait for balancing to quiesce...

Reproduced during `rolling-upgrade-with-sla-no-shares-test` It looks like it happens right after the raft topology upgrade procedure. ``` [2025-02-23T15:20:58.554Z] < t:2025-02-23 15:20:58,185 f:file_logger.py l:101 c:sdcm.sct_events.file_logger p:INFO > 2025-02-23 15:20:58.182: (InfoEvent Severity.NORMAL) period_type=not-set...

https://argus.scylladb.com/tests/scylla-cluster-tests/4f55ebf5-cddc-4f2e-ae92-d684978cdd8c replicated duplicate entries in col8 not pasting the error as it's too big wall of text, but basically ```diff --- expected_col8 +++ actual_col8 @@ -1,9 +1,18 @@ 5d606058a51d2f50... +5d606058a51d2f50......

Reproduced in 2024.2.5 > 2025-02-23 21:37:33.861 : (ScyllaBenchLogEvent Severity.ERROR) period_type=one-time event_id=f108a018-38e3-40f6-b042-46160c7539c2 during_nemesis=DecommissionStreamingErr: type=ConsistencyError regex=received only line_number=23156 node=Node longevity-large-partitions-200k-pks-loader-node-733161df-0-1 [34.139.122.119 | 10.142.15.243] 2025/02/23 21:37:33 [query statement="SELECT pk, ck, v FROM scylla_bench.test...

Found an interesting issue https://github.com/scylladb/scylla-cluster-tests/issues/10727 Some tests had 1 rack per dc even with 3 AZs defined ``` availability_zone: 'a,b,c' n_db_nodes: '6 6' ```

Tested decommissioning 2 nodes x 3 racks https://argus.scylladb.com/tests/scylla-cluster-tests/174be212-b9ac-4fc7-ae20-8b659bfb53ec  Old sequential 

> @tgrabiec claimed that some business logic changes would be required after the rebase. Please elaborate. I tried to cherry-pick this commit into the branch with out of space changes,...