Python批量刪除mysql中千萬(wàn)級(jí)大量數(shù)據(jù)
場(chǎng)景描述
線上mysql
數(shù)據(jù)庫(kù)里面有張表保存有每天的統(tǒng)計(jì)結(jié)果,每天有1千多萬(wàn)條,這是我們意想不到的,統(tǒng)計(jì)結(jié)果咋有這么多。運(yùn)維找過(guò)來(lái),磁盤占了200G
,最后問(wèn)了運(yùn)營(yíng),可以只保留最近3天的,前面的數(shù)據(jù),只能刪了。刪,怎么刪?
因?yàn)檫@是線上數(shù)據(jù)庫(kù),里面存放有很多其它數(shù)據(jù)表,如果直接刪除這張表的數(shù)據(jù),肯定不行,可能會(huì)對(duì)其它表有影響。嘗試每次只刪除一天的數(shù)據(jù),還是卡頓的厲害,沒(méi)辦法,寫個(gè)Python腳本批量刪除吧。
具體思路是:
- 每次只刪除一天的數(shù)據(jù);
- 刪除一天的數(shù)據(jù),每次刪除50000條;
- 一天的數(shù)據(jù)刪除完,開始刪除下一天的數(shù)據(jù);
Python代碼
# -*-coding:utf-8 -*-
import sys
# 這是我們內(nèi)部封裝的Python Module
sys.path.append('/var/lib/hadoop-hdfs/scripts/python_module2')
import keguang.commons as commons
import keguang.timedef as timedef
import keguang.sql.mysqlclient as mysql
def run(starttime, endtime, regx):
tb_name = 'statistic_ad_image_final_count'
days = timedef.getDays(starttime,endtime,regx)
# 遍歷刪除所有天的數(shù)據(jù)
for day in days:
print '%s 數(shù)據(jù)刪除開始'%(day)
mclient = getConn()
sql = '''
select 1 from %s where date = '%s' limit 1
'''%(tb_name, day)
print sql
result = mclient.query(sql)
# 如果查詢到了這一天的數(shù)據(jù),繼續(xù)刪除
while result is not ():
sql = 'delete from %s where date = "%s" limit 50000'%(tb_name, day)
print sql
mclient.execute(sql)
sql = '''
select 1 from %s where date = '%s' limit 1
'''%(tb_name, day)
print sql
result = mclient.query(sql)
print '%s 數(shù)據(jù)刪除完成'%(day)
mclient.close()
# 返回mysql 連接
def getConn():
return mysql.MysqlClient(host = '0.0.0.0', user = 'test', passwd = 'test', db= 'statistic')
if __name__ == '__main__':
regx = '%Y-%m-%d'
yesday = timedef.getYes(regx, -1)
starttime = '2019-08-17'
endtime ='2019-08-30'
run(starttime, endtime, regx)
循環(huán)判斷數(shù)據(jù),如果有,繼續(xù)刪除當(dāng)天50000
條數(shù)據(jù);否則,開始刪除下一天的數(shù)據(jù)。花了半個(gè)小時(shí),終于刪除完了。
作者:柯廣的網(wǎng)絡(luò)日志 ? Python批量刪除mysql中千萬(wàn)級(jí)大量數(shù)據(jù)
微信公眾號(hào):Java大數(shù)據(jù)與數(shù)據(jù)倉(cāng)庫(kù)