[SERVER-31998] Perf regression with mongodump integration tests Created: 16/Nov/17 Updated: 27/Oct/23 Resolved: 05/Jan/18 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | 3.6.0-rc4 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | David Golden | Assignee: | Backlog - Storage Execution Team |
| Resolution: | Works as Designed | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
darwin Sierra, linux Ubuntu 17.10 |
||
| Attachments: |
|
| Assigned Teams: |
Storage Execution
|
| Operating System: | ALL |
| Steps To Reproduce: |
|
| Participants: |
| Description |
|
When running the mongodump integration tests, I noticed a significant performance difference between 3.4.10 replica sets (WT) and 3.6.0-rc4 replica sets (both WT and MMAPv1). On my MBP, the difference was 50% to 100%+. On my 20-core desktop, the difference was ~ 18%. The workloads involved are tiny. |
| Comments |
| Comment by Ian Whalen (Inactive) [ 05/Jan/18 ] |
|
Thanks for filing this David. We're closing as Works as Designed because: 1) We're just doing more work in the drop/create code paths given the two-phase drops. |
| Comment by Mathias Stearn [ 14/Dec/17 ] |
|
My initial profiling indicated that this workload spends the majority of its time dropping and creating databases. Pushing to storage team to decide how/if we want to address this. |
| Comment by David Golden [ 16/Nov/17 ] |
|
Added output for the two configurations as tested on Ubuntu 17.10. Each file shows the command line parameters for the nodes and the timing results of two runs. |