[SERVER-43851] Work around zombie writes in $merge tests Created: 04/Oct/19  Updated: 06/Dec/22  Resolved: 12/Feb/21

Status: Closed
Project: Core Server
Component/s: Aggregation Framework, Testing Infrastructure
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Charlie Swanson Assignee: Backlog - Query Optimization
Resolution: Won't Fix Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
depends on SERVER-44252 Delete implicit collection creation l... Closed
Related
related to SERVER-38852 Failing $merge can leave zombie write... Backlog
Assigned Teams:
Query Optimization
Operating System: ALL
Participants:
Linked BF Score: 37

 Description   

SERVER-38852 describes an issue which will be difficult to fix but is causing some test failures. We should work around those test failures in the meantime.



 Comments   
Comment by Charlie Swanson [ 12/Feb/21 ]

After chatting with the QO team, we decided to close this as "Won't Fix", because the associated BFs seem to have died off. We will re-open and reconsider should that situation change.

Comment by Charlie Swanson [ 08/Nov/19 ]

So - plot twist. The 'allowImplicitCollectionCreation: false' flag we now set on $merge does not actually prevent the collection from being created. It just prevents the shard from doing it implicitly itself - it causes the config server to drive the collection create. This means that my workaround in these tests doesn't work. Fortunately, nicholas.zolnierz, janna.golden and esha.maharishi and I discussed and recognized that SERVER-44252 was planned to remove the flag for allowImplicitCollectionCreation. Instead of removing it, we plan to let that ticket proceed and take away any callers of allowImplicitCollectionCreation: false except for $merge. Once we do that, we can change what the flag means to better suit our purpose: under no circumstances is the collection to be created.

So for now I'm abandoning the code review I sent anton.korshunov and nicholas.zolnierz and I'm going to mark this as "depends on" SERVER-44252. This means that we won't be able to fix the BFs on the v4.2 branch, so I'll eventually mark them as trivial once we fix them on master.

Comment by Charlie Swanson [ 31/Oct/19 ]

Didn't get any real work done on this last week due to interviews.

Comment by Charlie Swanson [ 04/Oct/19 ]

I've got a proposed fix for BF-14122, but next week I'd like to look into how we can work around the other linked BFs. 

Generated at Thu Feb 08 05:04:17 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.