$match gets absorbed into ixscan when preceded by $group on $meta with accumulator on $meta

XMLWordPrintableJSON

    • Type: Bug
    • Resolution: Works as Designed
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • None
    • Query Integration
    • ALL
    • Hide

      1. Run a timeseries aggregation with the pipeline 

      pipeline = [{$group: {_id: "$meta", x: {$min: "$meta"}}}, {$match: {_id: 1}}]

       2. Print out the result of the optimized pipeline 

      coll.explain().aggregate(pipeline)
      Show
      1. Run a timeseries aggregation with the pipeline  pipeline = [{$group: {_id: "$meta" , x: {$min: "$meta" }}}, {$match: {_id: 1}}]  2. Print out the result of the optimized pipeline  coll.explain().aggregate(pipeline)
    • QI 2023-09-04
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      In a timeseries pipeline, when we first have a $group stage on the $meta field and an accumulator on the $min as the first stage, and then have a $match on the _id as the second stage, the pipeline is rewritten to absorb the $match into the index scan and leave the $group stage. 

      In essence, the stages $_unpack -> $group(_id=$meta, min=$meta) -> $match (_id=1) gets reordered to $IXSCAN (with $meta=1) -> $group(_id=$meta, min=$meta)

       

            Assignee:
            Gil Alon
            Reporter:
            Alice Li (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: