[SERVER-37680] Tests that use arbiters in replica sets and create majority commit point lag should disable majority read concern Created: 19/Oct/18 Updated: 27/Oct/23 Resolved: 08/May/19 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Testing Infrastructure |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Maria van Keulen | Assignee: | Maria van Keulen |
| Resolution: | Gone away | Votes: | 0 |
| Labels: | nyc | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Sprint: | Storage NYC 2018-12-03, Storage NYC 2019-05-20 | ||||||||
| Participants: | |||||||||
| Description |
|
The replica set flow control project aims to keep majority committed replication lag bounded. Tests that set up replica sets with arbiters, e.g., apply_batch_only_goes_forward.js may time out when the flow control mechanism is enabled. The flow control mechanism is intentionally disabled when majority read concern is disabled. Enabling majority read concern with a PSA replica set is not a supported configuration, so disable majority read concern in tests that use PSA replica sets to avoid timeouts. |
| Comments |
| Comment by Maria van Keulen [ 08/May/19 ] |
|
As an update, it looks like since this ticket was filed, offending tests likeĀ apply_batch_only_goes_forward.js no longer time out with Flow Control enabled. Closing as Gone Away. If another such timeout surfaces, the test can be addressed appropriately. |
| Comment by Maria van Keulen [ 14/Dec/18 ] |
|
Update: PSA replica sets with majority read concern enabled are no longer a supported configuration. |
| Comment by Maria van Keulen [ 26/Oct/18 ] |
|
Per a conversation with tess.avitabile, we would still like to have testing coverage for PSA replica sets and majority read concern enabled, since we expect these configurations to work properly, but be less performant when majority commit point lag is high. I am updating the work for this ticket accordingly. |
| Comment by Sara Williamson [ 26/Oct/18 ] |
|
maria.vankeulen Need to check with replication for impact on test coverage for majority reads on replica sets with arbiters. We're leaving this in needs triage for now. |