[SERVER-78259] Revise strong lock acquisitions on the chunk migrations code path Created: 20/Jun/23  Updated: 12/Dec/23

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

Type: Bug Priority: Major - P3
Reporter: Kaloian Manassiev Assignee: Nandini Bhartiya
Resolution: Unresolved Votes: 0
Labels: cs-subteam1, sharding-nyc-subteam1
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Assigned Teams:
Sharding NYC
Operating System: ALL
Participants:
Case:

 Description   

As part of the work to support distributed transactions (under SERVER-38147) we decided to cap the donor strong lock acquisitions to 500ms by default in order to avoid shard-wide stalls due to strong locks being stuck behind active transactions (which can possibly have longer duration). The consideration is that while customers have control of when to issue a DDL operation which may take a strong lock, they have no control of when the balancer decides to move chunk.

This ticket is to consider the same for the recipient side of chunk migration. Specifically we have the following acquisitions:
Prior to 4.4:

  • Indexes reconciliation between donor and recipient here
  • Updating the range deleter state here

After 4.4 only the first one remains.


Generated at Thu Feb 08 06:37:52 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.