Currently using mongodb timeseries collections to store IoT-Data. Data is then retrieved via multiple REST-APIs. In most cases those APIs fetching the last entry of a specified meta-data field.

Unfortunately as the time-series collections grow bigger (10 collections with around 4 mil entries each), im getting a “Slow Query” warning in the db logs and queries take unreasonably long (> 10 seconds) to return some value. Currently NO secondary index is setup.

My query (Golang code) looks like this

func (mh *MongoHandler) FindLast(collection string, nodeName string, exEmpty bool) ([]TimeSeriesData, error) {

    coll := mh.client.Database(mh.database).Collection(collection)
    filter := bson.D{
        {Key: "meta.nodeName", Value: nodeName},
    }

    if exEmpty {
        filter = append(filter, primitive.E{Key: "value", Value: bson.D{
            {Key: "$exists", Value: true},
            {Key: "$ne", Value: ""},
        }})
    }

    sortParams := bson.D{{Key: "ts", Value: -1}}

    var res []TimeSeriesData

    cursor, err := coll.Find(ctx, filter, options.Find().SetSort(sortParams), options.Find().SetLimit(1))

    if err != nil {
        return nil, err
    }
    cursor.All(ctx, &res)
    return res, nil
}

Can you help me to improve this query and speed it up? Would a secondary index on the timestamp field help me here?