[SERVER-48239] Move potentially expensive operations out of WT shutdown Created: 15/May/20  Updated: 06/Dec/22

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

Type: Improvement Priority: Major - P3
Reporter: Bruce Lucas (Inactive) Assignee: Backlog - Storage Engines Team
Resolution: Unresolved Votes: 0
Labels: KP44
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Assigned Teams:
Storage Engines
Participants:
Story Points: 5

 Description   

Potentially expensive operations are done during WT shutdown, but by that point gathering storage engine statistics for ftdc is no longer possible, making diagnosis of problems encountered during shutdown in the field difficult or impossible.

Since we already trigger the normal checkpoints in the integration layer, could we also do the final checkpoint there?

That's certainly possible, but it's a non-trivial architectural change. I'd be hesitant to make it at this stage in the 4.4 release but would be happy to explore in depth after the 4.4.0 release.



 Comments   
Comment by Sulabh Mahajan [ 11/Jun/21 ]

Note: The goal here is to be able to reduce the duration of FTDC gaps seen during the shutdown (and startup?). This helps continue debugging the system if the shutdown itself takes much longer. The investigation should keep that as its primary purpose and consider moving checkpoint or other processes outside of the shutdown as one of the possible means to do so. Could also investigate if we can continue providing FTDC well into the shutdown.

Comment by Deepti Hasija [ 10/Jun/21 ]

Scope: Investigate only. If we decide to go with the fix, create a separate WT ticket for an actual fix.

sulabh.mahajan, to add more context.

Generated at Thu Feb 08 05:16:33 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.