子文档上的猫鼬唯一索引

假设我有一个简单的架构:

Let's say I have a simple schema:

var testSchema = new mongoose.Schema({
    map: { type: [ mongoose.Schema.Types.Mixed ], default: [] },
    ...possibly something else
});

现在让我们确保对 (_id, map._id) 是唯一的.

Now let's ensure that pairs (_id, map._id) are unique.

testSchema.index({ _id: 1, 'map._id': 1 }, { unique: true });

使用 db.test.getIndexes() 快速检查显示它已创建.

Quick check using db.test.getIndexes() shows that it was created.

{
    "v" : 1,
    "unique" : true,
    "key" : {
        "_id" : 1,
        "map._id" : 1
    },
    "name" : "_id_1_map._id_1",
    "ns" : "test.test",
    "background" : true,
    "safe" : null
}

问题是，这个索引被忽略了，我可以轻松地创建多个具有相同 map._id 的子文档.我可以轻松地多次执行以下查询:

The problem is, this index is ignored and I can easily create multiple subdocuments with the same map._id. I can easily execute following query multiple times:

db.maps.update({ _id: ObjectId("some valid id") }, { $push: { map: { '_id': 'asd' } } });

并最终得到以下结果:

{
    "_id": ObjectId("some valid id"),
    "map": [
        {
            "_id": "asd" 
        },
        {
            "_id": "asd" 
        },
        {
            "_id": "asd" 
        }
    ]
}

这是怎么回事?为什么我可以推送冲突的子文档?

What's going on here? Why can I push conflicting subdocuments?