[SERVER-29261] test performance of initial sync when doing _id index scan vs $natural collection scan Created: 17/May/17  Updated: 06/Dec/22  Resolved: 10/Apr/18

Status: Closed
Project: Core Server
Component/s: Replication
Affects Version/s: None
Fix Version/s: None

Type: Task Priority: Major - P3
Reporter: Spencer Brody (Inactive) Assignee: Backlog - Replication Team
Resolution: Won't Fix Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Assigned Teams:
Replication
Participants:

 Comments   
Comment by Judah Schvimer [ 24/May/17 ]

Further testing did not show any regressions. We will attempt this test again once PERF-943 is resolved.

Comment by Judah Schvimer [ 22/May/17 ]

The first patch above showed no significant regressions, though also the LogKeeper test was broken, and one of the workload phases was broken. The normal initial sync workload is likely not a good indicator of this performance change, since the database can keep all of the data in memory and using an _id sort will have the largest impact when data is not in memory.

I'm rerunning the patch here with some of the workload bugs fixed: https://evergreen.mongodb.com/version/5922f74a2fbabe1b4d000242

Comment by Judah Schvimer [ 17/May/17 ]

I have run the following patch adding a "sort: {_id: 1}" to the collection cloner's find command:
https://evergreen.mongodb.com/version/591cb28f2fbabe24b50032bf

Generated at Thu Feb 08 04:20:20 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.