[SERVER-15152] When evaluating plans, some index candidates cause complete index scan Created: 05/Sep/14  Updated: 28/Dec/17  Resolved: 18/Sep/14

Status: Closed
Project: Core Server
Component/s: Querying
Affects Version/s: 2.6.4
Fix Version/s: 2.6.5, 2.7.7

Type: Bug Priority: Major - P3
Reporter: Asya Kamsky Assignee: David Storch
Resolution: Done Votes: 2
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Related
related to SERVER-14311 skipping of index keys is not account... Closed
is related to SERVER-14525 Perf regression in 2.6.2 caused by no... Closed
is related to SERVER-15528 Distinct queries can scan many index ... Closed
Tested
Operating System: ALL
Backport Completed:
Participants:
Case:

 Description   

When running query during multiPlanRunner stage some combinations of values and data distribute can cause a fast query have a possible plan/index choice where full index scan (of a non-selective attribute) would be evaluated. Running explain(true) can show which index has unusually high nscanned.

It seems during race-off the full scan is finished, causing some percentage of queries to be excessively slow.



 Comments   
Comment by Githook User [ 18/Sep/14 ]

Author:

{u'username': u'dstorch', u'name': u'David Storch', u'email': u'david.storch@10gen.com'}

Message: SERVER-15152 refactor IXSCAN stage to avoid doing too much in one work() cycle
Branch: v2.6
https://github.com/mongodb/mongo/commit/c7ee53ba9bf48dd958baa4cbc45ec5d6f3683072

Comment by David Storch [ 18/Sep/14 ]

Author:

{u'username': u'dstorch', u'name': u'David Storch', u'email': u'david.storch@10gen.com'}

Message: SERVER-15125 refactor IXSCAN stage to avoid doing too much in one work() cycle
Branch: master
https://github.com/mongodb/mongo/commit/a7cdcee52937fec949eea4050973bde5c1d825be

Generated at Thu Feb 08 03:37:06 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.