监控redis多实例的负载情况

唐伯虎

单个服务器上创建多实例，对其重要参数的监控是非常重要的，以下是监控服务器上多实例的负载情况：主要包含：redis实例的QPS、内存使用情况、客户端连接数，服务器的内存使用率、CPU使用率、CPU load值、网卡流量等，脚本把采集到的数据显示并且写入到数据库中，方便查看多个服务器的多实例情况以及数据的汇总等，写的优点仓促，有兴趣的同学可以整理一下脚本使其模块化、简洁化。脚本如下：

#!/usr/bin/env python--coding:UTF-8--
import os,threading,time,sys,sigar,MySQLdb
'''
安装python的sigar模块
apt-get install libtool automake gettext python-MySQLdb screen pkg-config git
git clone git://github.com/hyperic/sigar.git sigar.git
./autogen.sh
./configure
make
make install
cd bindings/python/
python setup.py install
'''
'''
建库sql
CREATE TABLE redis_stats (
 id int(11) NOT NULL AUTO_INCREMENT,
 host_name varchar(50) NOT NULL,
 qps int(11) NOT NULL,
 clients int(11) NOT NULL,
 redis_mem_used varchar(50) NOT NULL,
 sys_mem_used_pers float NOT NULL,
 cpu_used float NOT NULL,
 cpu_load varchar(50) NOT NULL,
 netband varchar(255) NOT NULL,
 uptime timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
PRIMARY KEY (id),
KEY host_name (host_name),
KEY uptime (uptime)
) ENGINE=InnoDB DEFAULT CHARSET=utf8
查询每个实例的最新记录
select host_name,qps,clients,redis_mem_used,concat(sys_mem_used_pers,'%') as sys_mem_used,concat(cpu_used,'%') as cpu_used,cpu_load,netband from DBA.redis_stats group by host_name desc order by qps desc;
'''
def log2db(check_log):
log_host = '192.168.56.101'
log_user = 'root'
log_pass = '1q2w3e4r'
try:
    conn = MySQLdb.connect(host = log_host,port = 3306,user = log_user,passwd = log_pass,charset='utf8',connect_timeout=20)
    cursor = conn.cursor()
    #cursor.execute(insert_sql)
    cursor.executemany("INSERT INTO redis.redis_stats (`host_name`,`qps`,`clients`,`redis_mem_used`,`sys_mem_used_pers`,`cpu_used`,`cpu_load`,`netband`) VALUES (%s,%s,%s,%s,%s,%s,%s,%s)",tuple(check_log))
    conn.commit()
    cursor.close()
    conn.close()
  except Exception,e:
    print e

　　def redis_info(host,port,res):
　　var = []
　　var.append(host)
　　var.append(port)
　　aaa = os.popen("redis-cli -h %s -p %s info|grep -v '#'|tr -s '\r\n'" % (host,port)).readlines()
　　dirc = {}
　　for i in aaa:

if i != '\r\n':
      a = i.strip()
      aa = a.split(":")
      dirc[aa[0]]=aa[1]
  var.append(dirc["connected_clients"])
  var.append(dirc["instantaneous_ops_per_sec"])
  var.append(dirc["used_memory_human"])
  res.append(var)

　　def main():
　　netband = {}
　　stime = 5
　　while True:

try:
      sg = sigar.open()
      mem = sg.mem()#内存
      mem_percent = "%.2f" % mem.used_percent()
      cpu = sg.cpu()#CPU总的使用率
      cpu_idle = "%.2f" % ((1-float(cpu.idle())/cpu.total())*100)
      loadavg = sg.loadavg()#CPU load值
      cpu_loadavg = ','.join([str(i) for i in loadavg])
      #nets = [i for i in sg.net_interface_list() if 'dum' not in i and i != 'lo']#网卡流量统计
      nets = [i.strip() for i in os.popen("/bin/ip a|grep global|awk '{print $7}'").readlines() if i.strip() != '']
      if len(netband) != 0:
        for net in nets:
          netband[net+'_Out'] = "%.2f" % (float((sg.net_interface_stat(net).tx_bytes()) - int(netband[net+'_Out']))/stime/1024/1024)
          netband[net+'_In'] = "%.2f" % (float((sg.net_interface_stat(net).rx_bytes()) - int(netband[net+'_In']))/stime/1024/1024)
      else:
        for net in nets:
          netband[net+'_Out'] = "%.2f" % (float(sg.net_interface_stat(net).tx_bytes())/stime/1024/1024)
          netband[net+'_In'] = "%.2f" % (float(sg.net_interface_stat(net).rx_bytes())/stime/1024/1024)
      redis_list = ['192.168.56.101:6379','192.168.1.87:16379']
      text = "*"*20 + "  Redis Status  %s  " % time.strftime("%Y-%m-%d %H:%M:%S") + "*"*20
      print "\033[1;31;40m%s\033[0m" % text
      threads = []
      res = []
      for i in redis_list:
        aa = i.split(':')
        host = aa[0]
        port = aa[1]
        t = threading.Thread(target=redis_info,args=(host,port,res))
        threads.append(t)
      for i in range(len(threads)):
        threads[i].start()
      for i in range(len(threads)):
        threads[i].join()
      print "\033[1;35;40m%s\033[0m" % ("Redis_host:port".ljust(23)+"Redis:QPS".ljust(10)+"Redis:Clients".ljust(15)+"Redis:Mem_used".ljust(15)+"Mem_percent".ljust(12)+"Cpu_used".ljust(10)+"Cpu_loadavg".ljust(17))
      All_qps = 0
      All_clients = 0
      res.sort()
      check_log = []
      for i in res:
        log = [i[0]+':'+i[1],int(i[3]),int(i[2]),i[4],float(mem_percent),float(cpu_idle),cpu_loadavg,str(netband)]
        check_log.append(log)
        print (i[0]+':'+i[1]).ljust(23)+i[3].ljust(10)+i[2].ljust(15)+i[4].ljust(15)+mem_percent.ljust(12)+cpu_idle.ljust(10)+cpu_loadavg.ljust(17)+str(netband)
        All_qps = All_qps + int(i[3])
        All_clients = All_clients + int(i[2])
      log2db(check_log)
      print "\033[1;35;40m%s\033[0m" % ("Summary All host:".ljust(23)+str(All_qps).ljust(10)+str(All_clients).ljust(10))
      netband = {}
      for net in nets:
        netband[net+'_Out'] = sg.net_interface_stat(net).tx_bytes()
        netband[net+'_In'] = sg.net_interface_stat(net).rx_bytes()
      time.sleep(stime)
    except KeyboardInterrupt :
      sys.exit(0)
      print
      break

　　if __name__ == "__main__":
main()有图有真相：