[SERVER-44525] Test that reads survive stepdowns more often with awaitable isMaster Created: 08/Nov/19  Updated: 12/Jun/23  Resolved: 12/Jun/23

Status: Closed
Project: Core Server
Component/s: Replication
Affects Version/s: None
Fix Version/s: 4.3 Desired

Type: Task Priority: Major - P3
Reporter: A. Jesse Jiryu Davis Assignee: [DO NOT USE] Backlog - Sharding NYC
Resolution: Won't Fix Votes: 0
Labels: sharding-common-backlog
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Assigned Teams:
Sharding NYC
Sprint: Service arch 2020-04-20, Service arch 2020-05-04, Service arch 2020-05-18
Participants:

 Description   

Compare the old protocol (clients send no maxAwaitTimeMS nor topologyVersion and don't set exhaustAllowed) with the new protocol (awaitable isMaster plus exhaust). We already have a workload that tests the performance of primary required reads in the face of stepdowns. Verify that a higher percentage of reads succeed with the new protocol, and verify that total throughput is higher. The test does not directly report the client time to recovery, but according to David Daly, "The underlying tool reports individual operation latencies, so we can figure out the exact recovery time after stepdown, although that may require some extra post-processing." Do this if we judge it worth the effort.



 Comments   
Comment by Lamont Nelson [ 10/Jul/20 ]

Moving this test to be a genny workload.

Comment by Lamont Nelson [ 09/Jul/20 ]

This is the code review for the jstest version of this test: https://mongodbcr.appspot.com/616380001

After talking with max.hirschhorn we decided to move this to a perf workload instead of jstest.

Comment by Tess Avitabile (Inactive) [ 23/Jan/20 ]

lamont.nelson, do you think the Sharding team can take on this work?

Comment by A. Jesse Jiryu Davis [ 23/Jan/20 ]

Agreed, sounds good to me.

Comment by Tess Avitabile (Inactive) [ 23/Jan/20 ]

jesse, I believe this test would require a client that implements the new protocol, is that correct? If so, I'll ask the Sharding team if they can take this on, or we can schedule it after the mongos implementation is done.

Generated at Thu Feb 08 05:06:13 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.