MongoDB Aggregation Limit Lookup

Starting from MongoDB 3.6 you can use uncorrelated subqueries to limit the lookup:

db.indicators.aggregate([{ $lookup: {  from: 'comments',  as: 'match_docs',  let: { indicator_id: '$_id' },  pipeline: [    { $match: {      $expr: { $eq: [ '$obj_id', '$$indicator_id' ] }    } },    // { $sort: { createdAt: 1 } }, // add sort if needed (for example, if you want first 100 comments by creation date)    { $limit: 100 }  ]} }])

I was able to figure it out.

$lookup -> $match -> $project

db.indicators.aggregate([{    "$lookup": {        "from": "comments"        , "localField": "_id"        , "foreignField": "obj_id"        , "as": "match_docs"    }}, {    "$match": {        "match_docs": {            "$exists": True        }    }}, {    "$project": {        "match_docs_agg": {            "$slice": ["$match_docs", 3]        }    }}])

If you do a $unwind immediately following a $lookup, the pipeline will be optimized, basically combining the 2 stages helping to bypass the 16MB limit that could result from the $lookup returning a large number of documents.

Keep in mind, if a single document in the foreign collection plus the size of the document in the local collection exceed 16 MB, this optimization cannot help.