[SERVER-21998] insert into capped with mmapv1 should be throttled (core_workloads_MMAPv1 failed with multiple node replSet) Created: 10/Dec/15  Updated: 06/Dec/22  Resolved: 14/Sep/18

Status: Closed
Project: Core Server
Component/s: MMAPv1, Storage
Affects Version/s: 3.2.0
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Rui Zhang (Inactive) Assignee: Backlog - Storage Execution Team
Resolution: Won't Fix Votes: 0
Labels: sys-perf-reg
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Assigned Teams:
Storage Execution
Operating System: ALL
Sprint: Performance E (01/11/16)
Participants:

 Description   
  • with workload to insert into capped collection with MMAPv1, the secondary could not keep up
  • the speed to drain oplog is slow with capped collection, this may cause performance test timeout for waiting for w:ALL (w:NUM_OF_NODES) write (2 hours)
  • the repro step is same as SERVER-21646, except test this with MMAPv1
  • this issue recently surfaced due to we added forceSyncOplog for some of the workload, thus triggered timeout.

for comparison, after fix for SERVER-21646, WR insert_capped operates at 12,230 ops, and MMAPv1 is doing 34,250 ops



 Comments   
Comment by Rui Zhang (Inactive) [ 15/Dec/15 ]

After test, this is issue with insert_capped, may related to SERVER-21646.

This is exposed by recent additional of w:ALL write

Generated at Thu Feb 08 03:59:06 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.