[SERVER-68689] Switch background validation to use checkpoint cursors again Created: 09/Aug/22  Updated: 29/Oct/23  Resolved: 10/Oct/22

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 6.2.0-rc0

Type: Task Priority: Major - P3
Reporter: Yuhong Zhang Assignee: Yuhong Zhang
Resolution: Fixed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
depends on SERVER-69135 Bring back the kCheckpoint ReadSource... Closed
depends on SERVER-69186 Track newly created indexes for backg... Closed
depends on SERVER-69247 Add concurrency control for taking ch... Closed
depends on WT-5927 Make checkpoint cursors work again Closed
Problem/Incident
Related
related to SERVER-46287 Switch background validation to use r... Closed
Backwards Compatibility: Fully Compatible
Sprint: Execution Team 2022-08-22, Execution Team 2022-09-05, Execution Team 2022-09-19, Execution Team 2022-10-17
Participants:
Linked BF Score: 145

 Comments   
Comment by Etienne Petrel [ 11/Oct/22 ]

I don't think it is creating noise at the moment, yuhong.zhang@mongodb.com, let us know if this changes. We are now working on WT-9877.

Comment by Keith Smith [ 11/Oct/22 ]

WT-9877 shouldn't be hard to fix. Based on the comments, the fix I suggested in this comment should work and it's a one or two line change. The hard part, IMHO, will be writing a reproducer. 

In terms of next steps, Storage Engines should prioritize the fix. As to whether to re-revert SERVER-68689, I think that depends on how much noise it's creating in testing. 

Comment by Etienne Petrel [ 11/Oct/22 ]

Thanks for the additional information yuhong.zhang@mongodb.com.
keith.smith@mongodb.com, vamsi.krishna@mongodb.com, should we ask for the revert of SERVER-68689 until WT-9877 is resolved? The good thing is that checkpoint cursor is not a feature available to customers yet and having more testing before it happens is definitely a great thing. However, we may generate more BF, potentially duplicates of BF-26388.

Comment by Yuhong Zhang [ 11/Oct/22 ]

Hi etienne.petrel@mongodb.com, unfortunately we were not aware of BF-26388 or WT-9877. The revert and push were due to another bug that checkpoint_timestamp in wiredtiger is not updated atomically along with the checkpoint operation itself. Do you think we should revert it again?

Comment by Etienne Petrel [ 11/Oct/22 ]

Hi yuhong.zhang@mongodb.com, I would like to confirm something.
We have WT-9877 open after this BF-26388 was created. It seems that we should fix WT-9877 before checkpoint cursors are used again.
However, I can see that SERVER-68689 was pushed, reverted and pushed. Is it safe or are we exposing ourselves to new occurrences of BF-26388?

Comment by Githook User [ 06/Oct/22 ]

Author:

{'name': 'Yuhong Zhang', 'email': 'yuhong.zhang@mongodb.com', 'username': 'YuhongZhang98'}

Message: SERVER-68689 Switch background validation to use checkpoint cursors again
Branch: master
https://github.com/mongodb/mongo/commit/9179d0a7109efb34631a6e19a87e055ed4e8c730

Comment by Githook User [ 19/Sep/22 ]

Author:

{'name': 'Yuhong Zhang', 'email': 'yuhong.zhang@mongodb.com', 'username': 'YuhongZhang98'}

Message: SERVER-68689 Temporarily disable background validation to use checkpoint cursors

Reverts the part of the implementation of the checkpoint cursors
Branch: master
https://github.com/mongodb/mongo/commit/f5324f56bfd410822c1ddad76515fa4cc9ea21e0

Comment by Githook User [ 06/Sep/22 ]

Author:

{'name': 'Yuhong Zhang', 'email': 'yuhong.zhang@mongodb.com', 'username': 'YuhongZhang98'}

Message: SERVER-68689 Switch background validation to use checkpoint cursors again
Branch: master
https://github.com/mongodb/mongo/commit/1a8f5ae47b2cd941b96cf8795fa45a167ac4fbeb

Generated at Thu Feb 08 06:11:28 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.