I am connecting to MongoDB Atlas using mongodb-spark-connector v10.3
I am using a SamplePartitioner with a pipeline that matches on a certain date range and projects only the _id column:

[{'$match': {'dateReceived':{'$gte': '2024-07-12T05:00:00.000000Z','$lte': '2024-07-12T09:15:00.000000Z'}}},{$project:{_id:1}}]

Since I am using the SamplePartitioner, the mongodb-spark-connector automatically appends the _id column to the $match stage and the resultant pipeline becomes:

[
         {
            "$match":{
               "$and":[
                  {
                     "_id":{
"$gte":"ca8866b865271263919bb74cb355cdf5955de54d80123368c9409ad12d622bcd",                        "$lt":"d039be1aaf69134b5b24009ca41d5763090bb8b77bd923b5ae3b38c6624d28c9"
                     }
                  },
                  {
                     "dateReceived":{
                        "$gte":"2024-07-12T05:00:00.000000Z",
                        "$lte":"2024-07-12T09:15:00.000000Z"
                     }
                  }
               ]
            }
         },
         {
            "$project":{
               "_id":true
            }
         }

The issue with this as I see, is the _id boundaries are created based on the samples it collects from each partition but it does not employ a min and a max range to get all the possible records.

Is my understanding correct?
Has anyone encountered this issue where the data returned from Mongo is not complete.