golang连接mysql时一般会加上loc=Local参数,否则timestamp字段储存和读取时间会有偏差,我们来看看为什么

timestamp原理

timestamp储存的是距离1970-01-01 00:00:00的秒数差,timestamp并不保存时区信息

准备表

CREATE TABLE `test`  (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `ts` timestamp(0) NULL DEFAULT NULL,
  PRIMARY KEY (`id`) USING BTREE
) ENGINE = InnoDB AUTO_INCREMENT = 1 ROW_FORMAT = Dynamic;

写入

insert into test(ts) values("2022-09-11 08:00:00");

插入上面数据时,数据库会将"2022-09-11 08:00:00"(北京时间)时间转成UTC时间"2022-09-11 00:00:00",再转成时间戳1662854400存入

select unix_timestamp(ts) from test;//返回1662854400

有些同学好奇了,数据库怎么知道"2022-09-11 08:00:00"这个时间是北京时间,原因是时区

SELECT @@GLOBAL.time_zone, @@SESSION.time_zone;//返回SYSTEM SYSTEM

global.time_zone: mysql服务设置的时区
session.time_zone: 此次连接的设置时区,一般就是global.time_zone,上面返回的SYSTEM,代表取系统时区,也就是东八区

Values for TIMESTAMP columns are converted from the session time zone to UTC for storage, and from UTC to the session time zone for retrieval.
value:"2022-09-11 08:00:00" session.time_zone:"东八区" => UTC "2022-09-11 00:00:00"=>1662854400

golang timestamp插入

有些同学可能以为连接参数loc=Local就是设置session.time_zone,没设置就会时区对不上,所以有问题,我们测试下

package main

import (
    "database/sql"
    "fmt"
    _ "github.com/go-sql-driver/mysql"
)

func main() {
    db, err := sql.Open("mysql", "root:xxx@tcp(127.0.0.1:3306)/test?parseTime=true")
    if err != nil {
        panic(err)
    }
    defer db.Close()

    var gZone, sZone string
    err = db.QueryRow("SELECT @@GLOBAL.time_zone, @@SESSION.time_zone").Scan(&gZone, &sZone)
    if err != nil {
        panic(err)
    }

    fmt.Println(gZone, sZone) //打印SYSTEM SYSTEM
}

可以看出,未设置loc=Local并不影响session.time_zone,它的值还是系统时区东八区

插入time.Time

清空数据,我们插入一条time.Time类型数据

db.Exec(`insert into test(ts) values (?)`, time.Now()) //北京时间2022-09-11 22:17:07

查下数据库数据

mysql> select ts,unix_timestamp(ts) from test;
+---------------------+--------------------+
| ts                  | unix_timestamp(ts) |
+---------------------+--------------------+
| 2022-09-11 14:17:07 |         1662877027 |
+---------------------+--------------------+
1 row in set (0.01 sec)

插入的是当前时间22:17:07,展示的却是14:17:07,我们再看比较下时间戳,北京时间2022-09-11 22:17:07的时间戳应为1662905827,但实际储存的是1662877027

(1662905827-1662877027)/3600 = 8//差了8小时

结论: 当未设置loc=Local时,timestamp储存time.Time确实有偏差

插入时间字符串

清空数据,我们插入一条时间字符串类型数据

db.Exec(`insert into test(ts) values (?)`, “2022-09-11 22:17:07”) //北京时间2022-09-11 22:17:07

查下数据库数据

mysql> select ts,unix_timestamp(ts) from test;
+---------------------+--------------------+
| ts                  | unix_timestamp(ts) |
+---------------------+--------------------+
| 2022-09-11 22:17:07 |         1662905827 |
+---------------------+--------------------+
1 row in set (0.02 sec)

时间字符串插入后数据没有偏差,是正确的

综上,只有向timestamp字段插入time.Time数据时,会有时间偏差

原因

还是看golang mysql库源码,mysql库的配置中有

func NewConfig() *Config {
    return &Config{
        Collation:            defaultCollation,
        Loc:                  time.UTC,
        MaxAllowedPacket:     defaultMaxAllowedPacket,
        AllowNativePasswords: true,
    }
}

未设置loc=Local时,Loc=time.UTC,当传入的参数是time.Time时:

        case time.Time:
               ...
                v := v.In(mc.cfg.Loc)
                v = v.Add(time.Nanosecond * 500) // To round under microsecond
                year := v.Year()
                year100 := year / 100
                year1 := year % 100
                month := v.Month()
                day := v.Day()
                hour := v.Hour() //当强制时区变换后,此Hour值会不同
                minute := v.Minute()
                second := v.Second()
                micro := v.Nanosecond() / 1000
              ...

强制时区变换,变换后的v.Hour()=14,测试代码如下

    t := time.Unix(1662905827, 0) //北京时间2022-09-11 22:17:07
    utc := t.In(time.UTC)

    fmt.Println(t.Hour(), ":", t.Minute(), ":", t.Second()) //打印 22:17:7
    fmt.Println(utc.Hour(), ":", utc.Minute(), ":", utc.Second())//打印 14:17:7

分析

db.Exec(`insert into test(ts) values (?)`, time.Now()) //北京时间2022-09-11 22:17:07
value:"2022-09-11 14:17:07" session.time_zone:"东八区" => UTC "2022-09-11 06:17:07"=>1662877027
db.Exec(`insert into test(ts) values (?)`, “2022-09-11 22:17:07”) //北京时间2022-09-11 22:17:07
value:"2022-09-11 22:17:07" session.time_zone:"东八区" => UTC "2022-09-11 14:17:07"=>1662905827

参照