[SERVER-81789] Don't kill already completed op in SessionWorkflow Created: 03/Oct/23  Updated: 01/Nov/23  Resolved: 01/Nov/23

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 7.2.0-rc0

Type: Improvement Priority: Major - P3
Reporter: Mathias Stearn Assignee: Erin McNulty
Resolution: Fixed Votes: 0
Labels: perf-tiger, perf-tiger-handoff, perf-tiger-triaged
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
is related to SERVER-81793 Don't do any syscalls to notify a bat... Closed
Assigned Teams:
Service Arch
Backwards Compatibility: Fully Compatible
Sprint: Service Arch 2023-10-30, Service Arch 2023-11-13
Participants:

 Description   

This code currently calls killAndDelistOperation() for an operation that has already completed. Instead it should just delist the operation. Killing the op is currently more expensive than it should be due to baton inefficiency (I'll file a linked ticket for that soon SERVER-81793), but even once that is fixed, it will still be cheaper not to kill the op.

This was measured to be the cause of a few percent of the findOne request latency.



 Comments   
Comment by Githook User [ 01/Nov/23 ]

Author:

{'name': 'Erin McNulty', 'email': 'erin.mcnulty@mongodb.com', 'username': 'erin2722'}

Message: SERVER-81789: Don't kill already completed op in SessionWorkflow
Branch: master
https://github.com/mongodb/mongo/commit/d525995aa9620d228817edeb3788e3ccec3eaa0c

Generated at Thu Feb 08 06:47:26 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.