[SERVER-11081] Disable top chunk split optimization for hashed sharding Created: 08/Oct/13  Updated: 05/Dec/22  Resolved: 23/Nov/22

Status: Closed
Project: Core Server
Component/s: Sharding
Affects Version/s: 2.4.6
Fix Version/s: None

Type: Improvement Priority: Major - P3
Reporter: Asya Kamsky Assignee: [DO NOT USE] Backlog - Sharding EMEA
Resolution: Done Votes: 0
Labels: RachitaD, ShardingRoughEdges, sharding-common-backlog
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Sharding EMEA
Participants:

 Description   

Using a hashed shard key, we should not be doing the 100/0 split on the maxKey range. Doing so means that we split it unevenly and thus its turn to split will happen again on the larger portion of it, and then when maxKey chunk fills again, it will again be split "unevenly".

Since a lot of the times balancing will be disabled with hashed shard key (since gross uneven distribution of chunks is not likely) you may end up with shard having maxKey chunk with very large number of chunks (though generally the same number of overall documents).


Generated at Thu Feb 08 03:24:50 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.