Redis (VII) - sentry

Redis（ 7、 ... and ）—— sentry

1、 brief introduction

sentry (sentinel) It's a Distributed systems , It's used to do... On each server in the master-slave structure monitor , In case of failure, through the voting mechanism choice new master And put all slave Connect to the new master.

2、 effect

• monitor
  • Keep checking master and slave Is it working . master Survival tests 、master And slave Operation detection
• notice （ remind ）
  • When there is a problem with the monitored server , To the other （ The sentry room , client ） Sending notice .
• Automatic failover
  • To break off master And slave Connect , Pick one slave As master, Will others slave Connect to the new master, And inform the client of the new server address

Be careful ：
The sentry is also a redis The server , It just doesn't provide data services Usually, the number of sentinels is singular

3、 Deploy sentinels

• Configure the master-slave structure of one drag two

• Three sentries （ The configuration is the same , Different ports ）

  • see sentinel.conf

• Activate the sentry

  redis-sentinel sentinel Port number  .conf
  

4、 working principle

Monitoring phase

• It is used to synchronize the status information of each node
  • Get each sentinel The state of （ Whether online ）
• obtain master The state of
  • master attribute
    • runid
    • role：master
    • each slave Details of
• Get all slave The state of （ according to master Medium slave Information ）
  • slave attribute
    • runid
    • role：slave
    • master_host、master_port
    • offset
    • …

[

Notification stage

• The Sentinels will get information synchronized with each other （ Information symmetry ）

Fail over

confirm master Offline

• When a sentry finds that the main server is down , Will master Medium SentinelRedistance Medium master Change it to SRI_S_DOWN（ Subjective offline ）, And inform the other sentinels , Tell them they found master Hang up .
• Other sentinels receive the information sent by the sentinel , Will also try to connect master, If more than half （ In the configuration file ） confirm master After hanging up , Will master Medium SentinelRedistance Medium master Change it to SRI_O_DOWN（ Objective offline ）

Select sentinels to deal with

• Upon confirmation of master After hanging up , A sentinel will be selected for failover （ It is up to the sentry to specify which slave To make a new master）.
• The filtering method is that sentinels send messages to each other , And vote , Those with more votes are elected .

Specific treatment

• The Sentinels chosen by him are responsible for the current slave Screening , The screening criteria are ：
  • Select the alternative from the server list master
  • online
  • Slow response
  • And the original master Disconnected for a long time
  • The principle of priority
    • priority
    • offset
    • runid
  • Send instructions （ sentinel ）
    • To the new master send out slaveof no one( Disconnect from the original master The connection of )
    • To the other slave send out slaveof new masterIP port （ Let others slave With the new master Connected to a ）