MySQL數據庫服務器安裝標準

目錄

• BIOS優化，陣列配置
• 操做系統基礎優化
• 安裝jemalloc內存管理器
• 安裝異步IO支持
• 網卡綁定軟中斷
• MySQL5.7安裝配置

---

BIOS優化，陣列配置

1. 關閉CPU節能，由於服務器品牌衆多，BIOS設置不相同，主要是關閉CPU節能，如C一、DELLR730已經智能設置，直接有個performance選項關閉了CPU節能。
2. 若是服務器是8塊硬盤，建議兩塊作RAID1裝系統，剩餘6塊作RAID10作數據分區，RAID1陣列緩存設置成 WriteThrough,RAID10設置成writeback,將有限陣列卡緩存給RAID10陣列用，若是是8塊以上的盤，組建兩組RAID10，兩組緩存策略都是writeback,一組裝系統以及存放順序IO類型的數據，好比redolog,歸檔日誌，mysql的binlog，一組作數據分區。
3. 使用XFS文件系統，數據分區用XFS文件系統，掛載參數用defaults,noatime,nodiratime,nobarrier，記住根分區是不能用這個掛載參數，否則你根分區下的目錄文件都沒有訪問時間，修改時間，只能用於數據庫文件分區。
4. 修改IO調度策略
   io調度器修改成deadline，若是是ssd或pcie-ssd設備，則修改成noop，下列兩種修改方式：
   • 在線動態修改，重啓失效

   echo "deadline" >> /sys/block/sdb/queue/scheduler
   #這裏sdb修改成實際的設備名稱，例如sda，或者sdc。

   • 永久生效修改/etc/grub.conf配置文件，在kernel那行最末尾加上 elevtor=deadline numa=off ，例如：

   vim /etc/grub.conf
   kernel /vmlinuz-2.6.32-504.el6.x86_64 ro root=UUID=af13b3dc-c142-42b7-8ed6-cb7c60608af2 rd_NO_LUKS  KEYBOARDTYPE=pc KEYTABLE=us rd_NO_MD crashkernel=auto LANG=zh_CN.UTF-8 rd_NO_LVM rd_NO_DM rhgb quiet elevator=deadline numa=off
   #要當前生效能夠這樣設置
   cat /sys/block/sda/queue/scheduler 
   echo deadline > /sys/block/sda/queue/scheduler

   • 覈實io調度器配置

   cat /sys/block/sda/queue/scheduler 
   noop anticipatory [deadline] cfq

5. 關閉numa
   新一代架構的NUMA不適用於跑數據庫的場景 。它本意是爲了提升內存利用率，但實際效果很差，反而可能致使一個CPU的內存尚有剩餘，但另外一個不夠用，發生SWAP的問題，所以建議直接關閉或者修改 NUMA的調度機制。
   • 修改/etc/grub.conf，關閉NUMA，重啓後生效

   #在kernel那行增長一個配置後重啓生效，例如： 
   kernel /vmlinuz-2.6.18-308.el5 ro root=LABEL=/1 elevator=deadline numa=off rhgb quiet

   • 修改/etc/init.d/mysqld或者mysqld_safe腳本，設定啓動mysqld進程時的NUMA調度機制，例如：

   numactl -- interleave=all /usr/bin/mysqld_safe --datadir="$datadir" --socket="$socketfile"  \
   --log-error="$errlogfile" --pid-file="$mypidfile" \
   --user=mysql >/dev/null 2>&1

操做系統基礎優化

1. 修改最大文件描述符，關閉selinux

   ulimit -SHn 65535
   echo "ulimit -SHn 65535" >> /etc/rc.local
   cat >>/etc/security/limits.conf<<EOF
   * soft nproc 102400
   * hard nproc 102400
   * soft nofile 102400
   * hard nofile 102400
   EOF
   
   sed -i 's/SELINUX=enforcing/SELINUX=disabled/g' /etc/selinux/config

2. 關閉不須要的服務，只留下crond，network，rsyslog，sshd ，sysstat，udev-post

3. 內核參數調整
   swappiness是Linux的一個內核參數，用來控制Linux物理RAM內存進行SWAP頁交換的相對權重，儘可能減小系統緩存被從內存中清除的狀況。取值範圍是0～100，vm.swappiness的值越低，Linux內核會盡可能不進行SWAP交換頁的操做，vm.swappiness的值越高，linux會越多的使用SWAP空間。Linux系統的默認值是60，當系統須要內存時，有60%的機率使用SWAP。對於大多數桌面系統設置爲100，能夠提升系統的總體性能；對於數據庫應用服務器設置爲0，能夠提升物理內存的使用率，進而提升數據庫服務響應性能。須要注意的是RHEL7以上版本，若是vm.swappiness = 0，有可能會因爲OOM問題，致使mysqld進程被OOM-Killer進程殺掉。

   減小swap的使用率，若是設置爲0，須要結合vm.overcommit_memory=2，能夠儘可能避免OOM，通常設置爲1

   vm.swappiness=1

   默認值是3797，保證物理內存有足夠空閒空間，防止突發性換頁

   vm.min_free_kbytes=204800

   默認是100，增大這個參數設置了虛擬內存回收directory和i-node緩衝的傾向，這個值越大。越易回收,儘可能保留可用內存

   vm.vfs_cache_pressure=150

   確保能持續將髒數據刷新到磁盤，避免瞬間I/O寫，產生嚴重等待和設置MySQL中的innodb_max_dirty_pages_pct低一點原理相似
   當文件系統緩存髒頁數量達到系統內存百分之多少時（默認10%）就會觸發pdflush/flush/kdmflush等後臺回寫進程運行，將必定緩存的髒頁異步地刷入硬盤

   vm.dirty_background_ratio=5

   當文件系統緩存髒頁數量達到系統內存百分之多少時（默認20%），系統會中止全部的應用層的IO寫操做，等待刷完數據後恢復IO。

   vm.dirty_ratio=10

   減小TIME_WAIT，提升TCP效率；

   net.ipv4.tcp_tw_recycle=1
   net.ipv4.tcp_tw_reuse=1

安裝jemalloc內存管理器

jemalloc內存分配方式與系統默認安裝的glic的malloc內存分配方式相比，能提升MySQL的性能，下降了系統CPU和內存資源的利用,關於這方便的壓測數據,請參考:http://www.linuxeye.com/Linux/1914.html
編譯安裝很簡單

wget http://www.canonware.com/download/jemalloc/jemalloc-3.6.0.tar.bz2
tar xvf jemalloc-3.6.0.tar.bz2
./configure
make && make install
#已方便安裝好mysql後使用，使用也很簡單在[mysqld_safe] 加上malloc-lib= /usr/local/lib/libjemalloc.so

安裝異步IO支持

爲了提升磁盤操做性能，當前的數據庫系統都採用異步IO（Asynchronous IO，AIO）的方式來處理磁盤操做。InnoDB存儲引擎亦是如此。
在InnoDB1.1.x以前，AIO的實現經過InnoDB存儲引擎中的代碼來模擬實現。而從InnoDB 1.1.x開始（InnoDB Plugin不支持），
提供了內核級別AIO的支持，稱爲Native AIO。所以在編譯或者運行該版本MySQL時，須要libaio庫的支持,centos最小化安裝默認是沒有安裝的,安裝也簡單:

yum install libaio-devel

網卡綁定軟中斷

網卡軟中斷不平衡，集中在一個CPU核心上（mpstat 查看%soft集中，一般是cpu0)，綁定軟中斷多個核心上,能夠用兩個腳原本綁定。本身喜歡用哪一個都行:

vim set_irq_affinity.py

#!/usr/bin/env python 
import re 
from os import system,popen 
from os import walk as walkdir 
from optparse import OptionParser 

RPS_CPUS_VALUE = 'ffffffff' 
RPS_FLOW_VALUE = '4096' 
RPS_RFS_DEFAULT = '0' 
interrupts_file = '/proc/interrupts' 
rps_cpus_list = [] 
rps_flow_list = [] 
#ENTRY_VALUE=32768 

def get_device(): 
return re.findall(r'([a-z]+\d+)\s+Link.*',popen('ifconfig').read()) 

def get_rfs_rps_file(net_device): 
rps_path = '/sys/class/net/' + net_device + '/queues/' 
for s in walkdir(rps_path): 
if len(s[2]) == 2: 
rps_cpus_list.append('/'.join([s[0],s[2][0]])) 
rps_flow_list.append('/'.join([s[0],s[2][1]])) 

def file_hander(TARGET,VALUE='0'): 
try: 
f_hander = open(TARGET,'w') 
f_hander.write(VALUE) 
finally: 
f_hander.close() 

def set_rfs_rps(net_device): 
get_rfs_rps_file(net_device) 
def set_rps_cpus_value(PATH): 
file_hander(PATH,RPS_CPUS_VALUE) 

def set_rps_flow_value(PATH): 
file_hander(PATH,RPS_FLOW_VALUE) 

map(set_rps_cpus_value,rps_cpus_list) 
map(set_rps_flow_value,rps_flow_list) 

def unset_rfs_rps(net_device): 
get_rfs_rps_file(net_device) 
def unset_rps_cpus_value(PATH): 
file_hander(PATH,RPS_RFS_DEFAULT) 

def unset_rps_flow_value(PATH): 
file_hander(PATH,RPS_RFS_DEFAULT) 

map(unset_rps_cpus_value,rps_cpus_list) 
map(unset_rps_flow_value,rps_flow_list) 

def set_irq_balance(): 
stop_irq_balance = 'service irqbalance stop' 
system(stop_irq_balance) 
interrupts_ct = open(interrupts_file) 
cores_nr = len(interrupts_ct.readline().split()) # 獲取CPU核心數 
irq_bit = 0 
while True: 
inter_line = interrupts_ct.readline() 
if inter_line == "": 
break 
js = inter_line.split() 
if len(js[-1]) > 5: 
if re.match(r'eth.-',js[-1][:5]): 
irq_nr = js[0][:-1] 
TARGET = '/proc/irq/%s/smp_affinity' %(irq_nr) 
VALUE = str(re.sub('0x','',hex(1 << irq_bit))) #1 << irq_bit 相對於2的N次方 ，hex() 二進制轉十六進制 
file_hander(TARGET,VALUE) 
irq_bit += 1 
if irq_bit == cores_nr: 
irq_bit = 0 

def unset_irq_balance(): 
start_irq_balance = 'service irqbalance start' 
system(start_irq_balance) 

def usage(): 
usage = '''================================================= 
Description: irq_balance_set && rfs_rps_set tools 
Usage: 
<script> -i : set irq smp_affinity 
-I : unset irq smp_affinity 
-r : set rfs && rps 
-R : unset rfs && rps 
'''
print usage 

if __name__ == '__main__': 

parser = OptionParser() 
parser.add_option("-i", action="store_true", 
dest="irq_true", 
default=False) 
parser.add_option("-I", action="store_true", 
dest="irq_false", 
default=False) 
parser.add_option("-r", action="store_true", 
dest="rps_true", 
default=False) 
parser.add_option("-R", action="store_true", 
dest="rps_false", 
default=False) 
(options, args) = parser.parse_args() 

if options.irq_true == True: 
set_irq_balance() 
print "irq_balance_set successfully" 
elif options.irq_false == True: 
unset_irq_balance() 
print "unset irq balance successfully" 
elif options.rps_true == True: 
device_list = get_device() 
map(set_rfs_rps,device_list) 
print "rfs&&rps configured successfully" 
elif options.rps_false == True: 
device_list = get_device() 
map(unset_rfs_rps,device_list) 
print "unconfigured rfs&&rps successfully" 
else: 
usage()

加入到開機啓動中

#!/bin/bash
CORE_SUM="`grep -c '^processor' /proc/cpuinfo`"
IRQ_SUM="`echo "1 2 4 8 10 20 40 80 100 200 400 800 1000 2000 4000 8000 10000 20000 40000 80000 100000 200000 400000 800000"| cut -d " " -f -${CORE_SUM}`"
IRQ_NUM="`echo ${IRQ_SUM}`"
for i in `grep -E '(eth[0-9]+|em[0-9]+)' /proc/interrupts | awk -F ":" '{print $1}' | sed 's/\ //g'`; do
        echo -e "${i}\t:`cat /proc/irq/${i}/smp_affinity`"
  
        y="`echo ${IRQ_NUM} | awk '{print $1}'`"
        echo ${y} > /proc/irq/${i}/smp_affinity
#echo "echo ${y} > /proc/irq/${i}/smp_affinity"
        if [ "${y}" == "`echo ${IRQ_SUM} | awk '{print $NF}'`" ]; then
                IRQ_NUM="`echo ${IRQ_SUM}`"
        else
                IRQ_NUM="`echo ${IRQ_NUM} | sed 's/^\([0-9]\+\)\ \(.*\)/\2/g'`"
        fi
  
        echo -e "----\t `cat /proc/irq/${i}/smp_affinity`"
done
######Enable RPS (Receive Packet Steering)
rfc=4096
cc=$(grep -c processor /proc/cpuinfo)
rsfe=$(echo $cc*$rfc | bc)
sysctl -w net.core.rps_sock_flow_entries=$rsfe
for fileRps in $(ls /sys/class/net/eth*/queues/rx-*/rps_cpus)
do
    echo fff > $fileRps
done
  
for fileRfc in $(ls /sys/class/net/eth*/queues/rx-*/rps_flow_cnt)
do
    echo $rfc > $fileRfc
done
  
tail /sys/class/net/eth*/queues/rx-*/{rps_cpus,rps_flow_cnt}

MySQL5.7安裝配置

建立用戶組

groupadd mysql
useradd -g mysql -s /sbin/nologin -M mysql

安裝並建立數據目錄

yum install wget gcc gcc-c++ make cmake ncurses-devel libtool zilib-devel openssl openssl-devel numactl -y
tar xf mysql-5.7.16-linux-glibc2.5-x86_64.tar.gz -C /usr/local
ln -sv /usr/local/mysql-5.7.16-linux-glibc2.5-x86_64 /usr/local/mysql
mkdir -p /data/mysql3306/{data,logs,tmp}
chown -R mysql:mysql /data/mysql3306/
chown -R mysql:mysql /usr/local/mysql/

建立/etc/my.cnf

#my.cnf
[client]
port            = 3306
socket          =  /data/mysql3306/data/mysql3306.sock
 
[mysql]
prompt="\\u@\\h:\\p [\\d]>"
#pager="less -i -n -S"
#tee=/data/mysql3306/query.log
no-auto-rehash
 
[mysqld]
#misc
user = mysql
basedir = /usr/local/mysql
datadir = /data/mysql3306/data
port = 3306
socket =  /data/mysql3306/tmp/mysql3306.sock
event_scheduler = 0
tmpdir =  /data/mysql3306/tmp
#timeout
interactive_timeout = 3600
wait_timeout = 3600
#character set
character-set-server = utf8
open_files_limit = 65535
max_connections = 500
max_connect_errors = 100000
lower_case_table_names =1
#symi replication
#rpl_semi_sync_master_enabled=1
#rpl_semi_sync_master_timeout=1000 # 1 second
#rpl_semi_sync_slave_enabled=1
#logs
log-output=file
slow_query_log = 1
slow_query_log_file = slow.log
log-error = error.log
log_warnings = 2
pid-file = mysql.pid
long_query_time = 1
#log-slow-admin-statements = 1
#log-queries-not-using-indexes = 1
log-slow-slave-statements = 1
#binlog
#binlog_format = STATEMENT
binlog_format = row
server-id = 13306
log-bin = /data/mysql3306/logs/mysql-bin
binlog_cache_size = 4M
max_binlog_size = 256M
max_binlog_cache_size = 1M
sync_binlog = 0
expire_logs_days = 10
#procedure
log_bin_trust_function_creators=1
#
gtid-mode = on
enforce-gtid-consistency=1
\#relay log
skip_slave_start = 1
max_relay_log_size = 128M
relay_log_purge = 1
relay_log_recovery = 1
relay-log=relay-bin
relay-log-index=relay-bin.index
log_slave_updates
#slave-skip-errors=1032,1053,1062
#skip-grant-tables
#buffers & cache
table_open_cache = 2048
table_definition_cache = 2048
table_open_cache = 2048
max_heap_table_size = 96M
sort_buffer_size = 128K
join_buffer_size = 128K
thread_cache_size = 200
query_cache_size = 0
query_cache_type = 0
query_cache_limit = 256K
query_cache_min_res_unit = 512
thread_stack = 192K
tmp_table_size = 96M
key_buffer_size = 8M
read_buffer_size = 2M
read_rnd_buffer_size = 16M
bulk_insert_buffer_size = 32M
#myisam
myisam_sort_buffer_size = 128M
myisam_max_sort_file_size = 1G
myisam_repair_threads = 1
#innodb
innodb_buffer_pool_size = 10G
innodb_buffer_pool_instances = 1
innodb_data_file_path = ibdata1:1024M:autoextend
innodb_flush_log_at_trx_commit = 2
innodb_log_buffer_size = 8M
innodb_log_file_size = 100M
innodb_log_files_in_group = 3
innodb_max_dirty_pages_pct = 50
innodb_file_per_table = 1
innodb_rollback_on_timeout
innodb_status_file = 1
innodb_io_capacity = 200
transaction_isolation = READ-COMMITTED
innodb_flush_method = O_DIRECT

5.7初始化

cd /usr/local/mysql
./bin/mysqld --initialize 
#或者
./bin/mysqld --initialize-insecure

5.7如下版本初始化

cd /usr/local/mysql
./scripts/mysql_install_db --defaults-file=/etc/my.cnf

安全加固

delete from mysql.user where user!='root' or host!='localhost';
truncate table mysql.db;
drop database test;
flush privileges;
cat /data/mysql3306/logs/error.log |grep "password"
2016-11-30T13:28:37.276714Z 1 [Note] A temporary password is generated for root@localhost: 6fT?TYzJI*Dg

環境變量

echo "export PATH=$PATH:/usr/local/mysql/bin" >> /etc/profile
soure /etc/profile

開機自啓

cp /usr/local/mysql/support-files/mysql.server /etc/init.d/mysql
chmod +x /etc/init.d/mysql
chkconfig add mysql
chkconfig mysql on