1 redis backup
It is recommended that rdb and aof be turned on simultaneously.rdb is on by default and aof is off by default.Refer to rdb and aof for additional details http://blog.csdn.net/guoxingege/article/details/48780745
requirepass 123456 ## The following is the rdb configuration #dbfilename: persisted data stored locally in a file dbfilename dump.rdb #dir: persisted data is stored at the local path, and if redis-cli is started under / redis/redis-3.0.6/src, the data is stored in the current SRC directory dir ./ ##When snapshot triggers, save <seconds> <changes> ##The following is 900 seconds before snapshot takes place with at least one change operation ##Caution is required in setting this value to assess how intensive changes are in the system ##You can turn off snapshot functionality by "save"" #save time, which means that one key has been changed for persistent storage at 900s intervals; 10 key300s have been changed for storage; and 10,000 key60s have been changed for storage. save 900 1 save 300 10 save 60 10000 ##Whether to block client's Change Operations when an error occurs in snapshot and cannot continue, Error may be due to disk full/disk failure/OS level exception, etc. stop-writes-on-bgsave-error yes ##Whether or not rdb file compression is enabled, defaulting to yes, and compression often means "extra cpu consumption", as well as this smaller file size and shorter network transfer time rdbcompression yes ##The following are aof configurations ##This option is the switch of aof function, default is "no", you can turn on aof function by "yes" ##Only under "yes" will features such as aof override/file synchronization take effect appendonly yes ##Specify aof file name appendfilename appendonly.aof ##There are three legal values for specifying the file synchronization policy in aof operations: always everysec no, which defaults to everysec appendfsync everysec ##Whether appendfsync suspends file synchronization during aof-rewrite,'no'means'not suspended','yes' means'suspended', and defaults to'no' no-appendfsync-on-rewrite no ##Minimum file size (mb,gb) triggered by aof file rewrite, rewrite will only be triggered if the aof file is larger than this size, default "64mb", recommended "512mb" auto-aof-rewrite-min-size 64mb ##The percentage of the aof file that should grow when this rewrite triggers, relative to the last rewrite. ##After each rewrite, redis records the size of the "new aof" file (for example, A) at this point, when the AOF file grows after A*(1 + p) ##The next rewrite is triggered, and each time an aof record is added, the size of the current aof file is detected. auto-aof-rewrite-percentage 100
1.2 Start the master server
docker run -d -p 6379:6379 \ --name=redis-master \ -v redis.conf:/etc/redis.conf redis:4.0.5 \ redis-server /etc/redis.conf
The primary node starts one.
1.3 Start from Server
docker run -d -p 6380:6379 \ --name=redis-slave \ --link=redis-master \ -v redis.conf:/etc/redis.conf redis:4.0.5 \ redis-server /etc/redis.conf --slaveof redis-master 6379
Start at least one from the node.
Create a redis master server, then create a redis slave node, specified by the redis-server - slaveof redis-master 6379 command, indicating that the current node service is a redis-master slave.
When data is manipulated on the redis primary node, the secondary node synchronizes the primary node data.
2 Create a redis-sentinel image
# Port on which the current Sentinel service is running port 26379 # Temporary folder used by Sentinel service runtime dir /tmp # Sentinel monitors a master redis instance named mymaster with an IP address of redis-master and port number of 6379, and at least $SENTINEL_is required to judge the master as invalidQUORUM Sentinel processes agree that automatic failover will not execute as long as the number of Sentinels agreed is not up to the standard sentinel monitor mymaster redis-master 6379 $SENTINEL_QUORUM #Specifies the number of milliseconds that Sentinel will need to invalidate the Redis instance.Sentinel marks the instance as subjectively offline when it does not return a PING or returns an error directly after that time.Only a Sentinel process marking an instance as subjectively offline does not necessarily result in automatic failover of the instance: only after a sufficient number of Sentinels have marked an instance as subjectively offline will the instance be marked as objectively offline, and automatic failover will take place sentinel down-after-milliseconds mymaster $SENTINEL_DOWN_AFTER # Specifies how many new primary instances can be synchronized from Redis instances at most when performing a failover. The smaller the number, the longer the synchronization time, and the longer the failover will take to complete when there are more Redis instances sentinel parallel-syncs mymaster 1 # If the failover operation is not completed within this time (ms), the failover is considered to have failed sentinel failover-timeout mymaster $SENTINEL_FAILOVER # Set Primary Service Password sentinel auth-pass mymaster $PASSWORD
Refer to other configurations http://download.redis.io/redis-stable/sentinel.conf
2.2 Start sentinel service
docker run -d -p 6379:6379 \ --name=redis-master \ -v redis.conf:/etc/sentinel.conf redis:4.0.5 \ redis-server /etc/sentinel.conf --sentinel
Sentinel nodes are configured at least two or more.SENTINEL_The number of QUORUM depends on the number of sentinel nodes, typically reduced by 1.
3 kubernetes script
For k8s related scripts, refer to:
[Deploy entinel-based highly available Redis clusters using Docker Compose] https://yq.aliyun.com/articles/57953
[redis persistence] http://blog.csdn.net/guoxingege/article/details/48780745