[SERVER-80107] Failovers can break time-series creation Created: 15/Aug/23 Updated: 22/Sep/23 Resolved: 22/Sep/23 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Felipe Gasper | Assignee: | Gregory Noma |
| Resolution: | Duplicate | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||
| Assigned Teams: |
Storage Execution NAMER
|
||||||||||||
| Operating System: | ALL | ||||||||||||
| Steps To Reproduce: | Unfortunately this isn’t easy to reproduce; it seems to have happened as a result of precise timing of the failover during the time-series collection’s creation. |
||||||||||||
| Sprint: | Execution NAMR Team 2023-10-02 | ||||||||||||
| Participants: | |||||||||||||
| Description |
|
In a recent test run against REP-2771’s development branch for time-series support, I saw this error:
The verbiage seems a bit frustrated, and there really shouldn’t be actual namespace conflicts at that point in a mongosync CI run since each namespace gets exactly 1 `create`, and nothing else alters the destination cluster at that time. It looks like what happened was that an election happened midway through creation of a time-series collection (via the public API, not creating buckets+view directly), and maybe that confused things afterward. One potential workaround is for mongosync to create the buckets & view separately; I haven’t yet investigated that change to see if it’ll work. |
| Comments |
| Comment by Felipe Gasper [ 22/Sep/23 ] |
|
gregory.noma@mongodb.com I’ve switched my mongosync branch to create buckets & view separately. So while this issue relates to REP-2771, it’s not blocking for us. |
| Comment by Gregory Noma [ 18/Aug/23 ] |
|
May be related to (or the same as) |