[SERVER-35191] Stuck with cache full during rollback Created: 23/May/18 Updated: 27/Oct/23 Resolved: 02/Oct/23 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Replication, WiredTiger |
| Affects Version/s: | 3.6.4 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Bruce Lucas (Inactive) | Assignee: | Backlog - Replication Team |
| Resolution: | Gone away | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Attachments: |
|
||||||||||||||||||||||||||||||||||||
| Issue Links: |
|
||||||||||||||||||||||||||||||||||||
| Assigned Teams: |
Replication
|
||||||||||||||||||||||||||||||||||||
| Operating System: | ALL | ||||||||||||||||||||||||||||||||||||
| Sprint: | Storage NYC 2018-09-10 | ||||||||||||||||||||||||||||||||||||
| Participants: | |||||||||||||||||||||||||||||||||||||
| Description |
|
During rollback we don't advance the oldest timestamp. This can pin a lot of data in the cache and we can get stuck with the cache full. |
| Comments |
| Comment by Lingzhi Deng [ 02/Oct/23 ] |
|
This problem goes away with WT history store |
| Comment by Ian Whalen (Inactive) [ 21/Aug/18 ] |
|
milkie believes that the important feature of triggering this issue is the stable timestamp lagging behind the majority commit point. Some combination of the mechanism in wt_unclean_shutdown.js to control checkpoints and RollbackTest to force operations to get rolled back in a 2-node + arbiter replica set should do the trick. |
| Comment by Ian Whalen (Inactive) [ 21/Aug/18 ] |
|
Reopening this so that Benety can address by adding a new test in case rollback and startup recovery have different behavior after prepare support is completed. Presumably no code changes. |
| Comment by Spencer Brody (Inactive) [ 23/May/18 ] |
|
This is likely effectively a duplicate of |