[SERVER-34839] Add a test to ensure writeConcern=majority waits are honored on unique index upgrade Created: 04/May/18 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: | Neha Khatri | Assignee: | Backlog - Storage Engines Team |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | nonnyc, storage-engines | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||
| Assigned Teams: |
Storage Engines
|
||||||||||||||||
| Sprint: | Storage Engines 2018-07-30 | ||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
Unique index upgrade waits for majority writes. Add test to ensure that majority write wait are honored. |
| Comments |
| Comment by Maria van Keulen [ 08/May/18 ] |
|
I think 4-5 days is an accurate estimate. I believe the test will involve setting a failpoint to cause a hang during the collMod that does the index upgrade and ensuring that the setFCV command times out, similar to the inserts in drop_collections_two_phase_write_concern.js. This test will be a bit involved, since both the setFCV command and the collMod writes wait for writeConcern, and ideally the test would specifically ensure the collMod writeConcern wait is honored. I asked the replication team for suggestions on how to do this and they suggested checking that no writes occurred after the collMod writeConcern wait. They pointed me to tests like migration_sets_fromMigrate_flag.js for examples on how to check writes after failpoints. |
| Comment by Alexander Gorrod [ 07/May/18 ] |
|
Complexity is unknown. Guess is 4-5 days. maria.vankeulen you asked for this test to be created as part of a code review, but we'd prefer to split the work out. Could you give us an idea of how long you think it might take to write the test? A pointer to a similar test - if it exists - would also be very helpful. |
| Comment by Alexander Gorrod [ 07/May/18 ] |
|
There is a ticket to capture the creation of a similar test for UUID work in |