MongoDB 日志切换(Rotate Log Files)实战
1. 在mongo shell下,执行logRotate命令:
use admin db.runCommand({logRotate:1})
需要在mongos,mongod,config server运行。
该方式的变种:
a) 在unix shell下运行:
mongo localhost/admin �Ceval “dbo.runCommand({logRotate:1})”
#!/bin/sh ### log rotate mongo localhost/admin �Cevel “db.runCommand({logRotate:1})” ### compress newly rotated for f in /var/log/mongodb/mongod.log.????-??-??T??-??-??; do 7za a “$f.z” “$f” rm �Cf “$f” done
c) 将如下脚本保存到logRotate.js文件:
db.getMongo().getDB(“admin”).runCommand({logRotate:1})
创建脚本logRotate.sh:
#!/bin/sh # Clear old logs rm /var/log/mongodb/mongod.log.* # Rotate logs mongo logRotate.js
d) logRotate.sh //写到计划任务crontab即可(需要expect软件包)
#!/usr/bin/expect �Cf spawn /usr/local/mongodb/bin/mongo admin -udev -ptest �Cquiet expect ">" send db.runCommand("logRotate") send "\r\n" expect ">" send "exit"
2. 使用SIGUSR1信号:
kill �CSIGUSR1 <mongod process id> find /var/log/mongodb/mongodb.log.* -mtime +7 �Cdelete
该方法的变种:
a) 用python写的定时脚本,每天产生一个新的log,超过7天的log自行删除。
#!/bin/env python import sys import os import commands import datetime,time #get mongo pid mongo_pid = commands.getoutput("/sbin/pidof mongod") print mongo_pid #send Sig to mongo if mongo_pid != '': cmd = "/bin/kill -USR1 %s" %(mongo_pid) print cmd mongo_rotate = commands.getoutput(cmd) else: print "mongod is not running..." #clean log which > 7 days str_now = time.strftime("%Y-%m-%d") dat_now = time.strptime(str_now,"%Y-%m-%d") array_dat_now = datetime.datetime(dat_now[0],dat_now[1],dat_now[2]) lns = commands.getoutput("/bin/ls --full-time /var/log/mongodb/|awk '{print $6, $9}'") for ln in lns.split('\n'): ws = ln.split() if len(ws) != 2: continue ws1 = time.strptime(ws[0],"%Y-%m-%d") ws2 = datetime.datetime(ws1[0],ws1[1],ws1[2]) if (array_dat_now - ws2).days > 7: v_del = commands.getoutput("/bin/rm -rf /var/log/mongodb//%s" % (ws[1]))
在root下crontab �Ce编辑定时任务
0 2 * * * /root/mongo_log_rotate.py >/root/null 2>&1
3. 日志管理工具logrotate
自动化的最好方式是使用logrotate,其中copytruncate参数能更好工作。
拷贝以下代码到/etc/logrotate.d/mongodb文件中,确保脚本中的路径和文件名正确。
# vi /etc/logrotate.d/mongodb /var/log/mongodb/*.log { daily rotate 7 compress dateext missingok notifempty sharedscripts copytruncate postrotate /bin/kill -SIGUSR1 `cat /var/lib/mongo/mongod.lock 2> /dev/null` 2> /dev/null || true endscript } # logrotate �Cf /etc/logrotate.d/mongodb
4. Mongodb bug
mongodb稳定性差强人意。在切换过程中也会导致mongodb进程终止。
具体内容可以查看下mongodb bug系统:SERVER-4739、SERVER-3339。