Details
-
Bug
-
Resolution: Duplicate
-
Major
-
7.6.2
-
Untriaged
-
0
-
Unknown
Description
Build:
7.6.0-2182(7.6.0-MP1)
Steps involved in the test
- Create 1 bucket with 30k items
- Create 2 fts index with 20 partitions
- Validate source partitions per index - validated
- rebalance in 2 nodes
- Validate source partitions per index - FAIL
Detailed failed step of validating source partitions
- It is balance check test(almost equal no of pindexes on all fts nodes)
- Claim is abs(num_node_partitions - exp_partitions_per_node) > partitions_per_pindex
- num_node_partitions is total sum of nodes_partitions[node]['pindexes'].items()
- Expected partitions per node is num_vbuckets/num_fts_nodes
- partitions_per_pindex can be found using `idx_defn['indexDef']['planParams']['maxPartitionsPerPIndex'])`
======================================================================
|
FAIL: rebalance_2_nodes_during_index_building (fts.vector_moving_topology_fts.VectorSearchMovingTopFTS)
|
----------------------------------------------------------------------
|
Traceback (most recent call last):
|
File "pytests/fts/vector_moving_topology_fts.py", line 2444, in rebalance_2_nodes_during_index_building |
self.is_index_partitioned_balanced(index)
|
File "pytests/fts/fts_base.py", line 5351, in is_index_partitioned_balanced |
% (num_node_partitions, node))
|
AssertionError: The source partitions are not evenly distributed among nodes, seeing 256 on c66eade50eb75a7314ccac36abc83710 |
|
----------------------------------------------------------------------
|
Attachments
Issue Links
- duplicates
-
MB-61043 Partition layout skew after failover(s) + rebalance; must not skip following rebalance ops in case of a skew
- Resolved