[SERVER-81355] Unclean shutdown after untimestamped collection drop can make catalog inconsistent. Created: 22/Sep/23  Updated: 27/Sep/23  Resolved: 27/Sep/23

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

Type: Bug Priority: Major - P3
Reporter: Suganthi Mani Assignee: Jordi Olivares Provencio
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Duplicate
duplicates SERVER-80974 Unclean shutdown while dropping local... Closed
Related
is related to SERVER-79245 Unclean shutdown while dropping colle... Closed
Assigned Teams:
Storage Execution EMEA
Operating System: ALL
Steps To Reproduce:

(Description to be filled)

Sprint: Execution EMEA Team 2023-10-02
Participants:
Linked BF Score: 144

 Description   

In our codebase, unclean node shutdowns after untimestamped collection drops can cause catalog inconsistencies (* idents known to _mdb_catalog but not to storage*), leading to node crashes on restart. This issue, observed in shard merge, is a general problem found in various areas of our codebase(see also SERVER-75740 which does untimestamped drop of system.profile collection).



 Comments   
Comment by Jordi Olivares Provencio [ 27/Sep/23 ]

SERVER-80974 is implementing a generic version of this algorithm that will solve this. It will make the case of an untimestamped delete of the catalog to wait until a checkpoint has been made before proceeding.

Generated at Thu Feb 08 06:46:16 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.