It's like haproxy except for redis.
Typically for every redis server we setup, we have a backup server setup as a slave of the main server.
If the Active Redis crashes or goes down for maintenance, we want the application to seamlessly use (read/write) data from the backup server. This can easily be done by making the backup server 'slaveof no one' But the problem is once the backup takes over as active new writes will go there. Now if the "original master" comes back up it will be out of sync with the "current master", so it has to be slaved to the "current master" before it is ready to start take requests.
This is solved by redis-proxy, which proxies the active redis. It is also smart enough to issue slave of commands to machines that start up and make masters slave of no one. It also can optionally provide for reads-write splitting. i.e. Reads going to slaves, and Writes only going to master.
This reduces the common redis slave master replication dance that needs to be done when bad stuff happens or maintenance of the servers are needed
Image courtesy Marlon Li(Marlon Li) represents how it is being or can be used.
The redis-proxy uses the config file to figure out the details of the redis cluster, The configurations are documented in the Config settings
We are in the process of testing it, it works for simple commands, but i have not tested and validated it against the whole set of redis commands. It is likely that commands like Pub sub don't work correctly(or at all).
Node.js and NPM are prerequistes. Here is the link to install both.
npm install -g redis-proxy
redis-proxy <path to config.json>
git clone firstname.lastname@example.org:sreeix/redis-proxy.git
Modify the config/config.json
The standard scenario is each redis has a backup redis.
If Both of them go down together, We just return errors and wait for one of them to come back on.
There can be only one master in the system. This is true even if multiple redis-proxies are monitoring the cluster.
There can be multiple slaves. Each will become slave of the master, and on master doing down, one of the slave it randomly picked as master.
Redis proxy can become a single point of failure, If it goes down your redis servers will become inaccessible. There are 2 possible setups
Using Nodemon/Forever to keep the redis proxy up all the time
Have a backup redis-proxy on Elastic IP or Virtual IP and switch manually or using keepalived. Note, however, that this is not a distributed system (like redis-sentinel). The choice of the redis server to be used is consistent. i.e if a proxy picks up the fact that master is down, then it would promote the highest slave (by runid) to master consistently.
Benchmarks show about 3x drop in performance(Investigating it and will post a fix soon)
No support for Monitoring & pub/sub commands( There is no reason why this can't be supported)
Would be nice to have a small ui for showing errors and status of redis servers.
It currently only works as master/slave mode. And it's highly unlikely that there could be a switch to sharded mode.
No downtime adding and removing slaves