利用数据库自定义并发 bunket 功能-白红宇

利用数据库自定义并发 bunket 功能

阅读量：4078 次

发布时间：2019-05-25

本文共 3103 字，大约阅读时间需要 10 分钟。

场景

防止由于停电后大规模 HV 启动, 导致 OPENSTACK NEUTRON SERVER 服务由于 HV 并发注册量太大而拒绝服务

思路

1. 数据库中创建表, COUNT, 列 id int 表中定义并发量, (例如 30)

2. 当 HV 启动时候, 假如需要启动 NEUTRON 服务, 则先要向 COUNT 表中获得信息,

假如 count.id 中值 > 1 , 则可以获得启动权限, 并 count.id -1,

当 neutron 服务注册完成, 则向数据库执行 count.id +1

参考表结构

用于做 bunket 算法create table count ( id int ) engine innodb;mysql> desc count;+-------+---------+------+-----+---------+-------+| Field | Type    | Null | Key | Default | Extra |+-------+---------+------+-----+---------+-------+| id    | int(11) | YES  |     | NULL    |       |+-------+---------+------+-----+---------+-------+1 row in set (0.00 sec)用于做数据记录的表, 可有可无CREATE TABLE node (  id int(11) DEFAULT NULL,  name varchar(50) DEFAULT NULL) ENGINE=InnoDB;mysql> desc node;+-------+-------------+------+-----+---------+-------+| Field | Type        | Null | Key | Default | Extra |+-------+-------------+------+-----+---------+-------+| id    | int(11)     | YES  |     | NULL    |       || name  | varchar(50) | YES  |     | NULL    |       |+-------+-------------+------+-----+---------+-------+2 rows in set (0.00 sec)

参考存储过程

获得 bunket 数据的存储过程

drop procedure startup;delimiter // create procedure startup( in d varchar(50), out n varchar(5) ) begin  declare a int;  declare b int;	declare c varchar(50);   select id into a from compute.count for update;   set autocommit=0;   if a > 0   then	   select name into c from compute.node where name=d;     if c is null     then       set b=a-1;       insert into compute.node values ( a, d);       update compute.count set id=b;       set n='ready';     else       set n='faile';     end if;   end if;   commit;   select n from dual; end //delimiter ;

完成neutron 注册后回调 bunket 的存储过程

drop procedure startend;delimiter // create procedure startend( in d varchar(50) ) begin  declare a int;  declare b int;	declare c varchar(50);   select id into a from compute.count for update;   set autocommit=0;   set b=a+1;   update compute.count set id=b;   delete from compute.node where name=d;   commit; end //delimiter ;

参加 compute 节点启动时候的启动脚本

#!/bin/bash##source params   <--- 注意:  公共变量,  公共函数存放的位置,  下文中缺的函数与变量都来自这里,  与整个算法无关, 忽略吧service_type=computesqlstartpath='/root/manager/get_status.sql'sqlendpath='/root/manager/end.sql'status='faile'if [ ! -d '/root/manager/' ]then    mkdir -p '/root/manager'ficat <
   
     $sqlstartpathset @a='ok';use compute;call startup('$hostname',@a);EOFcat <
    
      $sqlendpathuse compute;call startend('$hostname');EOFif [ ! -f "/usr/bin/mysql" ]then  yum install -y mysql > /dev/null 2>&1fiwhile [ "$status" != "ready" ]do  status=`mysql -h $sql_server -u $sql_user -p"$sql_passwd" --skip-column-names $sql_db -e "source $sqlstartpath;"`    if [ "$status" != "ready" ]    then        sleep 5    fidonestart_server messagebus messagebusstart_server libvirtd libvirtdstart_server openstack-nova-compute nova_computestart_server neutron-openvswitch-agent openvswitch_agentstart_server openvswitch openvswitchstart_server openstack-ceilometer-compute ceilometer_computemysql -h $sql_server -u $sql_user -p"$sql_passwd" --skip-column-names $sql_db -e "source $sqlendpath;"