Part1:写在最前
MongoDB的 unique index索引这里有个坑，主要体现在对NULL值的处理上，本文加以复现。
整体环境：
MongoDB 3.2.5
Part2:集合内容
PRIMARY> db.helei.find()
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcef5"), "i" : 0, "username" : "user0", "age" : 8, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcef6"), "i" : 1, "username" : "user1", "age" : 9, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcef7"), "i" : 2, "username" : "user2", "age" : 82, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcef8"), "i" : 3, "username" : "user3", "age" : 48, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcef9"), "i" : 4, "username" : "user4", "age" : 27, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcefa"), "i" : 5, "username" : "user5", "age" : 53, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcefb"), "i" : 6, "username" : "user6", "age" : 42, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcefc"), "i" : 7, "username" : "user7", "age" : 56, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcefd"), "i" : 8, "username" : "user8", "age" : 5, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b7ea9544e98b24a5bdcefe"), "i" : 9, "username" : "user9", "age" : 56, "created" : "Thu Mar 02 2017 17:49:09 GMT+0800 (CST)" }
{ "_id" : ObjectId("58b8da80d8509e8f46fd9042"), "i" : "10", "age" : 50, "create" : ISODate("2017-03-03T02:52:48.834Z") }
这里可以看到，我在helei集合中生成了i从0到10这些数据，而i:10文档中我故意漏写username这一个键值
Part3:集合索引情况
PRIMARY> db.helei.getIndexes()
[
{
"v" : 1,
"key" : {
"_id" : 1
},
"name" : "_id_",
"ns" : "helei.helei"
},
{
"v" : 1,
"key" : {
"age" : 1
},
"name" : "idx_age",
"ns" : "helei.helei"
},
{
"v" : 1,
"unique" : true,
"key" : {
"username" : 1
},
"name" : "uk_username",
"ns" : "helei.helei",
"background" : true
}
]
这里可以看出在name列添加了unique index： uk_username
Part4:验证
当再有一个不包含username键值的文档被插入时，会抛出错误
PRIMARY> db.helei.insert({i:"11",age:51,create:new Date()})
WriteResult({
"nInserted" : 0,
"writeError" : {
"code" : 11000,
"errmsg" : "E11000 duplicate key error collection: helei.helei index: uk_username dup key: { : null }"
}
})
Warning:警告
如果一个文档没有对应的键，索引会将其作为null存储。
——总结——
如果对某个键建立了唯一索引，但插入了多个缺少该索引键的文档，由于集合已经存在一个该索引键值的值为null而导致插入失败。由于笔者的水平有限，编写时间也很仓促，文中难免会出现一些错误或者不准确的地方，不妥之处恳请读者批评指正。
本文转自 dbapower 51CTO博客，原文链接：http://blog.51cto.com/suifu/1902878
，如需转载请自行联系原作者
