Mongo Map减少错误

时间:2013-04-12 10:03:00

标签: mongodb mapreduce

作为MongoDB的新手,我正在尝试进行MapReduce调用。 不幸的是我收到以下错误:

mongos> db.events.mapReduce(m,r,"errors");
Fri Apr 12 11:39:15.637 JavaScript execution failed: map reduce failed:{
    "ok" : 0,
    "errmsg" : "MR parallel processing failed: { errmsg: \"exception: reduce -> multiple not supported yet\", code: 10075, ok: 0.0 }"
} at src/mongo/shell/collection.js:L970
mongos> 

我正在使用Mongo 2.4.0

我的活动集合如下所示:

{ "sid" : "1UADM45MCGRTW ", "time" : ISODate("2013-04-08T11:33:13.229Z"), "class" : "S", "service" : "service1" }
{ "sid" : "1UADM45MCGRTW ", "time" : ISODate("2013-04-08T11:33:13.229Z"), "class" : "I", "service" : "service1" }
{ "sid" : "1UADM45MCGRTW ", "time" : ISODate("2013-04-08T11:33:13.236Z"), "class" : "E", "text" : "error message" }
{ "sid" : "1UADM45MCGRTW ", "time" : ISODate("2013-04-08T11:33:13.239Z"), "class" : "F", "service" : "service1" }

{ "sid" : "1UDO3H7YUOK08 ", "time" : ISODate("2013-04-08T11:33:08.095Z"), "class" : "S", "service" : "service2" }
{ "sid" : "1UDO3H7YUOK08 ", "time" : ISODate("2013-04-08T11:33:08.095Z"), "class" : "I", "service" : "service2" }
{ "sid" : "1UDO3H7YUOK08 ", "time" : ISODate("2013-04-08T11:33:08.173Z"), "class" : "E", "text" : "another error message" }
{ "sid" : "1UDO3H7YUOK08 ", "time" : ISODate("2013-04-08T11:33:08.188Z"), "class" : "F", "service" : "service2" }

{ "sid" : "1UDO3JVXIS2UZ ", "time" : ISODate("2013-04-08T11:28:39.480Z"), "class" : "I", "service" : "service1" }
{ ""sid" : "1UDO3JVXIS2UZ ", "time" : ISODate("2013-04-08T11:28:39.480Z"), "class" : "S", "service" : "service1" }
{ "sid" : "1UDO3JVXIS2UZ ", "time" : ISODate("2013-04-08T11:28:40.853Z"), "class" : "F", "service" : "service1" }
{ "sid" : "1UDO3JVXIS2UZ ", "time" : ISODate("2013-04-08T11:28:40.852Z"), "class" : "I", "service" : "service1" }

我的目的是获取所有“E”条目并将它们与“F”条目的相应“sid”的“服务”字段组合。所以这应该很简单,但我无法理解(因为上述错误)。

我使用的map / reduce函数是:

var m = function() {
    if (this.class == "E") {
        value = { time: this.time, error: this.text };
        emit(this.sid, value);
    } else if (this.class == "F") {
        value = { service: this.service };
        emit(this.sid, value);
    }
}


var r = function(key,values) {
    return values;
}

db.events.mapReduce(m,r,"errors");

2 个答案:

答案 0 :(得分:4)

问题是你想累积每个键的所有错误,这意味着你必须发出一个文档数组,然后你的reduce必须将它们组合成一个数组(因为它将接收一个数组数组) )。

你遇到的另一个问题是你想要将每个唯一cid的不同类组合成一些组合的简化结果,但你的reduce函数甚至都没有尝试这样做。

很难知道每个sid可能获得多少不同的类值,但如果它像你给出的示例数据那样,我猜你想要这样的东西:

var m = function() {
    array = [];
    if (this.class == "E") {
        value = { time: this.time, error: this.text };
        array.push(value);
        emit(this.sid, {a:array});
    } else if (this.class == "F") {
        value = { service: this.service };
        array.push(value);
        emit(this.sid, {a:array});
    }
}


var r = function(key,values) {
    result = {a:[]};
    values.forEach(function (v) {
       result.a = v.a.concat(result.a);
    });
    return result;
}

db.events.mapReduce(m,r,"errors");

在您的示例中,结果集合将如下所示:

> db.errors.find().pretty()
{
    "_id" : "1UADM45MCGRTW ",
    "value" : {
        "a" : [
            {
                "service" : "service1"
            },
            {
                "time" : ISODate("2013-04-08T11:33:13.236Z"),
                "error" : "error message"
            }
        ]
    }
}
{
    "_id" : "1UDO3H7YUOK08 ",
    "value" : {
        "a" : [
            {
                "service" : "service2"
            },
            {
                "time" : ISODate("2013-04-08T11:33:08.173Z"),
                "error" : "another error message"
            }
        ]
    }
}
{
    "_id" : "1UDO3JVXIS2UZ ",
    "value" : {
        "a" : [
            {
                "service" : "service1"
            }
        ]
    }
}

你可以添加一个finalize函数来将这个数组转换为一个单个文档的结果,但这取决于你想要的结果集合的格式以及每个sid的预期输入。

答案 1 :(得分:0)

根据文件(troubleshoot reduce function) “ reduce函数必须返回一个对象,其类型必须与map函数发出的值的类型相同

当您发出一个对象时,reduce也应该也会发出一个对象 - 所以将values包裹在一个对象中你应该很好。