[SERVER-45211] Sharded cluster fails on mapReduce with output collection when namespace does not yet exist Created: 17/Dec/19  Updated: 27/Oct/23  Resolved: 18/Dec/19

Status: Closed
Project: Core Server
Component/s: MapReduce
Affects Version/s: None
Fix Version/s: None

Type: Improvement Priority: Major - P3
Reporter: Jeffrey Yemin Assignee: Backlog - Query Team (Inactive)
Resolution: Works as Designed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
is related to JAVA-3555 Work around changes to mapReduce impl... Closed
Assigned Teams:
Query
Participants:

 Description   

A Java driver test of mapReduce to a non-existent database exposes a difference in behavior between latest builds and the 4.2 release. We're seeing this error:

{
   "ok": 0.0, 
   "errmsg": "database output-2284299633124 not found", 
   "code": 26, 
   "codeName": "NamespaceNotFound"

on latest sharded clusters. The test succeeds on all released versions of MongoDB.



 Comments   
Comment by James Wahlin [ 18/Dec/19 ]

This is expected behavior for the aggregation based mapReduce implementation and is inline with behavior for an aggregation pipeline with a $merge stage.

There are a number of behaviors supported by the legacy mapReduce implementation that do not fit in with our current sharded architecture (another example being mapReduce sharding the output collection as part of cluster command invocation). While removal of these features will be backward breaking for some users, they will provide a more stable / correct implementation.

Generated at Thu Feb 08 05:08:12 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.