[SERVER-55822] [v4.4 only] an index catalog entry may not have an ident on startup recovery when running with majority read concern off and repl two-phase drop Created: 06/Apr/21  Updated: 06/Dec/22  Resolved: 08/Apr/21

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

Type: Bug Priority: Major - P3
Reporter: Dianna Hohensee (Inactive) Assignee: Backlog - Storage Execution Team
Resolution: Won't Do Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Related
Assigned Teams:
Storage Execution
Operating System: ALL
Participants:
Linked BF Score: 0

 Description   

In v4.4 with majority read concern off, we use repl two-phase drop to rename a collection out of the way and then drop it when the drop timestamp is majority committed. If there is a checkpoint taken between starting an index build and dropping the collection+index wt tables for majority confirmed drop, startup recovery will have the index catalog entry and no index table. In standalone recovery mode, we skip restarting index builds, leaving the server in a state wherein we have an index catalog entry and no index table. Then, later, we try to instantiate a WiredTigerIndex instance on the non-existent index table.


Generated at Thu Feb 08 05:37:35 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.