Mongodb – How to interpret this “db.explain()” for this mongodb query

You can get overall (instance wide) hit info from the db.serverStatus() command, and in particular:

http://docs.mongodb.org/manual/reference/command/serverStatus/#serverStatus.indexCounters.hits http://docs.mongodb.org/manual/reference/command/serverStatus/#serverStatus.indexCounters.misses

These numbers are totals seen for the lifetime of the process, so if you want to get a rate or see the trend over time, then you will need to record them periodically and diff the values accordingly. Thankfully, if you install MMS Monitoring, it will do that for you and graph the results.

However, there are currently no per-index statistics available. The relevant feature request can be found here for tracking and voting purposes:

https://jira.mongodb.org/browse/SERVER-2227

Update: January 2016

The stats referenced above have been removed from the output of the server status command, so are no longer available. However, the referenced feature request is now complete and is available in version 3.2 in the form of the $indexStats aggregation operator (the linked docs also contain sample output). For completeness, here's an example I put together:

Prior to any queries I have just 2 indexes, the default _id and indexme, both with 0 ops:

> db.foo.aggregate( [ { $indexStats: { } } ] ).pretty()
{
    "name" : "indexme_1",
    "key" : {
        "indexme" : 1
    },
    "host" : "localhost:27017",
    "accesses" : {
        "ops" : NumberLong(0),
        "since" : ISODate("2016-01-12T19:03:01.358Z")
    }
}
{
    "name" : "_id_",
    "key" : {
        "_id" : 1
    },
    "host" : "localhost:27017",
    "accesses" : {
        "ops" : NumberLong(0),
        "since" : ISODate("2016-01-12T18:59:24.292Z")
    }
}

Then run a couple of finds to bump the ops on indexme and check again:

> db.timecheck.find({indexme: 33})
> db.timecheck.find({indexme: 55})

> db.timecheck.aggregate( [ { $indexStats: { } } ] ).pretty()
{
    "name" : "indexme_1",
    "key" : {
        "indexme" : 1
    },
    "host" : "localhost:27017",
    "accesses" : {
        "ops" : NumberLong(2),
        "since" : ISODate("2016-01-12T19:03:01.358Z")
    }
}
{
    "name" : "_id_",
    "key" : {
        "_id" : 1
    },
    "host" : "localhost:27017",
    "accesses" : {
        "ops" : NumberLong(0),
        "since" : ISODate("2016-01-12T18:59:24.292Z")
    }
}

First, 2.3 is a development branch that was turned into 2.4, you should not be using it any longer - hopefully you mean 2.2, which is still supported and developed (though it too will soon be end of life as of writing this answer).

The group method you mention runs server side javascript and is not going to be fast, especially in 2.2 which used the old spidermonkey engine and takes an exclusive javascript lock. Generally I would not recommend using it, especially with a sharded cluster, where it is unsupported.

Instead you should use the aggregation framework, which was added in 2.2, is improved in 2.4 and will be even better in 2.6. The framework includes a $group operator, and and many others that you can use in a "pipeline" to achieve the results you want. If that seems odd, then I recommend the pipeline explanation docs here - for anyone familiar with the Linux/Unix shell, the concept should be very familiar.

To do a group on your field you would do something like this (this would be more specific and easier to be specific if you had actually included a sample document and desired output):

db.mytbl.aggregate(
  [
    { $group : { _id : {date:"$date_col"} , count : { $sum : 1 } } }
  ]
)

You can find some more general examples here:

http://docs.mongodb.org/manual/applications/aggregation/