Mongodb aggregation - why is it so slow?

Question

PHP

Good evening everyone!
There is a limited collection:

db.createCollection("Error", {capped:true, size:419430400, max:200000})

The collection stores different types of errors from different servers.
It is necessary to count the number of errors of each type on each server in the last 24 hours.
My query option:

db.Error.aggregate (
    [
        {
            "$match" : {
                "dateAdd" : {
                    "$gt" : ISODate("2014-05-11T13:16:52Z"),
                    "$lte" : ISODate("2014-05-12T13:16:52Z")
                }
            }
        },
        {
            "$group" : {
                "_id" : {
                    "server" : "$server",
                    "errorType" : "$errorType"
                },
                "count" : {
                    "$sum" : NumberLong(1)
                }
            }
        }
    ]
)

When the collection is filled to the maximum (200000 errors), the request is executed in 0.5 seconds.
The index, judging by explain, is picked up.
Existing indexes:

db.Error.ensureIndex({web:1, hash:1, dateAdd:1});
db.Error.ensureIndex({type:1, dateAdd:1})
db.Error.ensureIndex({web:1, type:1, dateAdd:1});
db.Error.ensureIndex({dateAdd:1})

Answer 1

solution - a background task that runs aggregation once a day and saves the results in an add. collection.