membase自动化failover

    在membase web console台，可以对membase cluster中的server进行fail over,remove等操作，这些操作是手动进行，也可以通过membase提供的CLI接口自动进行。我设想这样一个环境，由两台server组成的membase cluster，server1 : 66.172.38.3 ，server2 : 66.172.38.4  .划分三个Bucket，分别对应11211,11212,11213，如web application连接该分布式缓存,如下示：66.172.38.3:11211,66.172.38.3:11212,66.172.38.3:11213， 在运行的过程中，server1 DOWN机，则web application会和membase失去联系。为避免这种情况，将moxi代理安装到web application端，使用localhost:11211,localhost:11212,localhost:11213连接membase,下面脚本可在两台server间提供failover.

 ###################################################################################

#!/bin/bash

status_code_3=`/usr/bin/curl -o /dev/null -s -w %{http_code} http://66.172.38.3:8091/index.html`
status_code_4=`/usr/bin/curl -o /dev/null -s -w %{http_code} http://66.172.38.4:8091/index.html`

function_bind_3()
{
/opt/moxi/bin/moxi -u root -Z port_listen=11211 http://66.172.38.3:8091/pools/default/bucketsStreaming/default &
/opt/moxi/bin/moxi -u root -Z port_listen=11212 http://66.172.38.3:8091/pools/default/bucketsStreaming/base2 &
/opt/moxi/bin/moxi -u root -Z port_listen=11213 http://66.172.38.3:8091/pools/default/bucketsStreaming/base3 &
}

function_bind_4()
{
/opt/moxi/bin/moxi -u root -Z port_listen=11211 http://66.172.38.4:8091/pools/default/bucketsStreaming/default &
/opt/moxi/bin/moxi -u root -Z port_listen=11212 http://66.172.38.4:8091/pools/default/bucketsStreaming/base2 &
/opt/moxi/bin/moxi -u root -Z port_listen=11213 http://66.172.38.4:8091/pools/default/bucketsStreaming/base3 &
}

function_failover_3()
{
failover_flag=`/opt/membase/bin/cli/membase server-list -c 66.172.38.4:8091  -u Administrator -p 123456 | grep 66.172.38.3 | awk '{print $3$4}'`

if [ x$failover_flag == "xunhealthyactive" ];then
echo "-----$(date)-------" >> /root/myselffile/membase/failover.log
/opt/membase/bin/cli/membase failover -c 66.172.38.4:8091 --server-failover=66.172.38.3:8091 -u Administrator -p 123456 >> /root/myselffile/membase/failover.log 2>&1
/opt/membase/bin/cli/membase rebalance -c 66.172.38.4:8091 -u Administrator -p 123456 >> /root/myselffile/membase/failover.log 2>&1
echo "-----End-----------" >> /root/myselffile/membase/failover.log
fi
}

function_failover_4()
{
failover_flag=`/opt/membase/bin/cli/membase server-list -c 66.172.38.3:8091  -u Administrator -p 123456 | grep 66.172.38.4 | awk '{print $3$4}'`

if [ x$failover_flag == "xunhealthyactive" ];then
echo "-----$(date)-------" >> /root/myselffile/membase/failover.log
/opt/membase/bin/cli/membase failover -c 66.172.38.3:8091 --server-failover=66.172.38.4:8091 -u Administrator -p 123456 >> /root/myselffile/membase/failover.log 2>&1
/opt/membase/bin/cli/membase rebalance -c 66.172.38.3:8091 -u Administrator -p 123456 >> /root/myselffile/membase/failover.log 2>&1
echo "-----End-----------" >> /root/myselffile/membase/failover.log
fi
}

if [ x$status_code_3 == "x200" ]; then
   function_bind_3
   server_flag="x3"
   else
     if [ x$status_code_4 == "x200" ]; then
     function_bind_4
     server_flag="x4"
     fi
fi

while true
do
status_code_3=`/usr/bin/curl -o /dev/null -s -w %{http_code} http://66.172.38.3:8091/index.html`
status_code_4=`/usr/bin/curl -o /dev/null -s -w %{http_code} http://66.172.38.4:8091/index.html`

if [ x$status_code_3 == "x000" ];then
   if [ x$server_flag == "xx3" ];then
      if [ x$status_code_4 == "x200" ];then
      killall moxi
      function_failover_3
      function_bind_4
      server_flag="x4"
      fi
   fi
fi

if [ x$status_code_4 == "x000" ];then
   if [ x$server_flag == "xx4" ];then
      if [ x$status_code_3 == "x200" ];then
      killall moxi
      function_failover_4
      function_bind_3
      server_flag="x3"
      fi
   fi
fi

if [ x$status_code_3 == "x000" ];then
   if [ x$server_flag == "xx4" ];then
      if [ x$status_code_4 == "x200" ];then
      function_failover_3
      fi
   fi
fi


if [ x$status_code_4 == "x000" ];then
   if [ x$server_flag == "xx3" ];then
      if [ x$status_code_3 == "x200" ];then
      function_failover_4
      fi
   fi
fi
done

 ####################################################################################
本环境只设定是两台server构成的membase cluster的failover情况,照此可以扩展为多台server构成的cluster的failover情况。