redis属于单进程的服务，它主要受内存、CPU、磁盘IO（主要是做持久化），如果服务器配置比较高，多核CPU、高内存的服务器，可以考虑做redis多实例。做多实例之前，首先要考虑CPU和内存的利用，我在测试的时候发现，redis在QPS为6-8W左右的时候，这个redis所在的逻辑CPU核的负载就在100%左右，所以要优化CPU使用这块，目前一般是是做网卡软中断来实现平衡这种单进程使用CPU过高的情况，不过需要网卡支持网卡软中断，效果不错。
多实例redis的管理，涉及到监控、服务的管理等，这里只说redis多实例的重启。以下是多实例的重启脚本，仅供交流参考：
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import redis,threading,sys,socket,time,os
from multiprocessing import Pool
def port_check(host,port):
POOL = redis.ConnectionPool(host=host, port=port)
my_server = redis.Redis(connection_pool=POOL)
try:
a = my_server.ping()
if a:
return 1
else:
return 0
except Exception,e:
return 0
def shutdown_server(host,port):
flag = port_check(host,port)
if flag == 1:
POOL = redis.ConnectionPool(host=host, port=port)
my_server = redis.Redis(connection_pool=POOL)
my_server.shutdown()
while 1:
flag = port_check(host,port)
if flag == 0:
print "%s:%s is Shutdown OK" % (host,port)
break
else:
time.sleep(1)
else:
print "%s:%s is Shutdown alreaday" % (host,port)
def start_server(host,port):
flag = port_check(host,port)
if flag == 0:
start_conm = "/usr/local/bin/redis-server /usr/local/redis/etc/redis_%s.conf" % port
os.popen(start_conm)
time.sleep(3)
i = 0
while 1:#每5s检测一次redis ping,持续一分钟，1分钟没有起来认为启动失败
flag = port_check(host,port)
if flag == 1:
print "%s:%s is Start OK" % (host,port)
break
else:
if i > 12:
print "%s:%s is Start Error" % (host,port)
break
else:
time.sleep(5)
i = i + 1
else:
print "%s:%s is Start alreaday" % (host,port)
def restart(host,port):
shutdown_server(host,port)
start_server(host,port)
def main():
server_list = ["127.0.0.1:6379","127.0.0.1:16379"]
pool = Pool(processes=3)
for i in server_list:
aa = i.split(':')
host = aa[0]
port = int(aa[1])
results = pool.apply_async(restart, (host,port))
pool.close()
pool.join()
if results.successful():
print 'successful'
if __name__=="__main__":
main()
©著作权归作者所有：来自51CTO博客作者AIOPS_DBA的原创作品，如需转载，请注明出处，否则将追究法律责任
重启脚本pythonNosql