MongoDB 计数很慢

Question

我有一个收藏产品，里面有 ~7.000.000 本书和总共 ~40GB mongodb 3.4 数据库。这是一本书文档的示例:

{ 
    "_id" : ObjectId("597f17d22be7925d9a056e82"), 
    "ean13" : "9783891491904", 
    "price" : NumberInt(2100), 
    "name" : "My cool title", 
    "author_name" : "Doe, John", 
    "warengruppe" : "HC", 
    "book_category_key" : "728",
    "keywords": ["fairy tale", "magic", "fantasy"]
    ...
}

当我用limit查询数据库时，时间是可以的。但是如果我算查询(用于分页)，那么它需要很长时间:

2017-08-02T13:03:16.088+0200 I COMMAND [conn74] 命令 mydb.products 命令:count { count: "products", query: { book_category_key: { $in: [ "120", "130", "180", "111", "112", "140", "150", "160", "170", "190", "1AA"] } }, readConcern: {} } planSummary: IXSCAN { book_category_key :1} keysExamined:1129826 docsExamined:1129825 numYields:8851 reslen:44锁:{全局:{acquireCount:{r:17704}}，数据库:{acquireCount:{r:8852}}，集合:{acquireCount:{r: 8852 } } } 协议(protocol):op_query 7008ms

这里是一个很好的查询:

{
    count: "products",
    query: {
        book_category_key: {
            $in: ["120",
            "130",
            "180",
            "111",
            "112",
            "140",
            "150",
            "160",
            "170",
            "190",
            "1AA"]
        }
}

这需要 7 秒，有时甚至更多(最多 20 秒)。我在 book_category_key 上有一个索引:

{ 
    "v" : 2, 
    "name" : "book_category_key_1", 
    "ns" : "mydb.products", 
    "background" : true
}

Answer 1

问题出在 planSummary: IXSCAN 上。当计数使用 IXSCAN 时，它也会进行 FETCH。像这样:

"planSummary" : "IXSCAN { book_category_key: 1 }",
"execStats" : {
    "stage" : "COUNT",
    ..... 
    "inputStage" : {
        "stage" : "FETCH",
        ....
        "inputStage" : {
            "stage" : "IXSCAN",
            .....

在您的案例中，它加载了您整个收藏的大约 1/7。

您可以投票给https://jira.mongodb.org/browse/SERVER-17266和相关问题，并使用建议的解决方法强制 COUNT_SCAN:

let cnt = 0;
for(let category of ["120",
    "130",
    "180",
    "111",
    "112",
    "140",
    "150",
    "160",
    "170",
    "190",
    "1AA"]) { cnt += db.g.count({book_category_key: category})};
print(cnt);

这是什么

"planSummary" : "COUNT_SCAN { book_category_key: 1 }",
"execStats" : {
    "stage" : "COUNT",
    ...
    "inputStage" : {
        "stage" : "COUNT_SCAN"
        ....

对于每个类别，如果索引适合内存，应该快约 10 倍。