[SERVER-68492] Test CE accuracy based on the JS golden-test framework Created: 02/Aug/22  Updated: 17/Feb/23  Resolved: 17/Feb/23

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

Type: Task Priority: Major - P3
Reporter: Timour Katchaounov Assignee: Backlog - Query Optimization
Resolution: Won't Do Votes: 0
Labels: M7
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Query Optimization
Participants:

 Description   

Test the accuracy of cardinality estimation via end-to-end tests against real data in mongod collections. The tests should be implemented via the golden-test framework developed by SERVER-67415.

This task should consider how to generate the dataset(s) used for the test. So far the most likely approach is to use the Python framework developed by WRITING-11332 to pre-generate datasets, and commit to master only the JS tests with their dataset(s).

In more detail:

  • generate a variety of datasets similar to the ones in the hist-ce tests,
  • evaluate and decide whether to pre-record the generated datasets, or generate them dynamically,
  • evaluate the cumulative accuracy results, and compare them to the hist-ce results,
  • record expected results using the golden test framework,
  • the testing framework should compute a baseline for each test either via running a separate count query, or explain("executionStats") and pick the actual counts from there.


 Comments   
Comment by Timour Katchaounov [ 17/Feb/23 ]

This task has been split into multiple sub-tasks implemented in the scope of milestone 7.

Generated at Thu Feb 08 06:10:57 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.