[SERVER-72618] Untimestamped writes to the catalog can occur when dropping unfinished indexes during replication rollback Created: 09/Jan/23  Updated: 29/Oct/23  Resolved: 18/Jan/23

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 6.3.0-rc0

Type: Bug Priority: Major - P3
Reporter: Gregory Wlodarek Assignee: Gregory Wlodarek
Resolution: Fixed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Problem/Incident
causes SERVER-82036 Relax startup recovery invariant for ... Closed
Related
related to SERVER-73858 Inconsistent error handling in singe-... Closed
Assigned Teams:
Storage Execution
Backwards Compatibility: Fully Compatible
Operating System: ALL
Sprint: Execution Team 2023-01-23
Participants:
Linked BF Score: 58

 Description   

During replication rollback, all index builds are stopped. When reopening the catalog, any unfinished index builds that weren't secondary background index builds have their index entry in the catalog dropped. This performs an untimestamped write. There's no oplog entry associated with this write, but we can probably use the common point timestamp the node rolled back to.



 Comments   
Comment by Githook User [ 18/Jan/23 ]

Author:

{'name': 'Gregory Wlodarek', 'email': 'gregory.wlodarek@mongodb.com', 'username': 'GWlodarek'}

Message: SERVER-72618 Oplog batch application index builds on an empty collection are completed within the same storage transaction
Branch: master
https://github.com/mongodb/mongo/commit/83fdd157932297d3e8dee3c4b0ff9d8014bbbe8e

Generated at Thu Feb 08 06:22:20 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.