[SERVER-78915] [only affect testing] A shard may be involved in more than one migration per balancer round when using the "Random migrations" feature Created: 13/Jul/23  Updated: 14/Jul/23  Resolved: 13/Jul/23

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: 5.0.19, 4.4.23, 6.0.8, 7.0.0-rc7
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Silvia Surroca Assignee: [DO NOT USE] Backlog - Sharding EMEA
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Duplicate
duplicates SERVER-78498 Make the balancer failpoint smarter Closed
Assigned Teams:
Sharding EMEA
Operating System: ALL
Participants:

 Description   

When the balancerShouldReturnRandomMigrations fail point is enabled, the balancer may schedule multiple migrations involving the same shards.

That is happening because the shards involved in a chosen migration are not removed from the availableShards list.

 



 Comments   
Comment by Silvia Surroca [ 13/Jul/23 ]

cris.insignarescuello@mongodb.com tommaso.tocci@mongodb.com sergi.mateo-bellido@mongodb.com I considered this a bug because this is not an expected behavior of the balancer.

I also wanted to give this ticket some visibility since it could be the root cause for still non-solved bugs, like for those BFs with migration failures that were closed as trivial because of the lack of coredumps and data files (BF-29107, BF-29140).

Generated at Thu Feb 08 06:39:37 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.