-
Type:
Bug
-
Resolution: Done
-
Priority:
Minor - P4
-
Affects Version/s: 3.0.4
-
Component/s: Aggregation Framework
-
Fully Compatible
-
ALL
-
Query F (02/01/16), Query 10 (02/22/16)
The SHARDING_FILTER stage seems to always be included in aggregations on shard servers, even if the connection is not versioned.
This is inconsistent with how find() works, which defies user expectations. It means that it's not possible to run an aggregation on orphaned documents (which would be strange, but could be desirable).
I'm not sure, but it may also mean that the presence of a migration might not be properly noticed by an in-progress agg. This will depend on how the shard versioning is handled, and I would much more strongly expect it in the case of an unversioned connection. This might actually cause results to be wrong (either by not noticing that some documents have moved away, or not noticing that more have arrived). In the case of an unversioned direct connection, this isn't entirely unexpected (since they normally see orphans anyway), but it might mean that aggregations in this case could sometimes avoid orphans, and sometimes not, which wouldn't be great.
Observed in 3.0.4.
mongos> sh.status()
--- Sharding Status ---
sharding version: {
"_id" : 1,
"minCompatibleVersion" : 5,
"currentVersion" : 6,
"clusterId" : ObjectId("55b749b3a7717d4ea1a796fa")
}
shards:
{ "_id" : "shard01", "host" : "shard01/genique:11112" }
{ "_id" : "shard02", "host" : "shard02/genique:11113" }
databases:
{ "_id" : "admin", "partitioned" : false, "primary" : "config" }
{ "_id" : "test", "partitioned" : true, "primary" : "shard01" }
test.test
shard key: { "_id" : 1 }
chunks:
shard02 1
shard01 1
{ "_id" : { "$minKey" : 1 } } -->> { "_id" : 0 } on : shard02 Timestamp(2, 0)
{ "_id" : 0 } -->> { "_id" : { "$maxKey" : 1 } } on : shard01 Timestamp(2, 1)
mongos> db.test.find()
{ "_id" : -2, "not_an_orphan" : 1 }
{ "_id" : -3, "not_an_orphan" : 2 }
mongos> db.test.aggregate([])
{ "_id" : -2, "not_an_orphan" : 1 }
{ "_id" : -3, "not_an_orphan" : 2 }
shard01:PRIMARY> db.test.find()
{ "_id" : -1 }
shard01:PRIMARY> db.test.aggregate([])
shard01:PRIMARY> db.test.find().explain(true)
{
"queryPlanner" : {
"plannerVersion" : 1,
"namespace" : "test.test",
"indexFilterSet" : false,
"parsedQuery" : {
"$and" : [ ]
},
"winningPlan" : {
"stage" : "COLLSCAN",
"filter" : {
"$and" : [ ]
},
"direction" : "forward"
},
"rejectedPlans" : [ ]
},
"executionStats" : {
"executionSuccess" : true,
"nReturned" : 1,
"executionTimeMillis" : 0,
"totalKeysExamined" : 0,
"totalDocsExamined" : 1,
"executionStages" : {
"stage" : "COLLSCAN",
"filter" : {
"$and" : [ ]
},
"nReturned" : 1,
"executionTimeMillisEstimate" : 0,
"works" : 3,
"advanced" : 1,
"needTime" : 1,
"needFetch" : 0,
"saveState" : 0,
"restoreState" : 0,
"isEOF" : 1,
"invalidates" : 0,
"direction" : "forward",
"docsExamined" : 1
},
"allPlansExecution" : [ ]
},
"serverInfo" : {
"host" : "genique",
"port" : 11112,
"version" : "3.0.4",
"gitVersion" : "0481c958daeb2969800511e7475dc66986fa9ed5"
}
}
shard01:PRIMARY> db.test.aggregate([], {explain:true})
{
"stages" : [
{
"$cursor" : {
"query" : {
},
"queryPlanner" : {
"plannerVersion" : 1,
"namespace" : "test.test",
"indexFilterSet" : false,
"parsedQuery" : {
"$and" : [ ]
},
"winningPlan" : {
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> "stage" : "SHARDING_FILTER",
"inputStage" : {
"stage" : "COLLSCAN",
"filter" : {
"$and" : [ ]
},
"direction" : "forward"
}
},
"rejectedPlans" : [ ]
}
}
}
],
"ok" : 1,
"$gleStats" : {
"lastOpTime" : Timestamp(1438078475, 1),
"electionId" : ObjectId("55b749b326a7d1afefbbfebb")
}
}