RCProxy
RCProxy 
Forked from https://github.com/wayslog/aster. Thanks!
======================
RCProxy is a lightweight, fast but powerful Redis Cluster Proxy written in Rust.
It supports memcache/redis singleton/redis cluster protocol all in one. RCProxy can proxy with two models:
- proxy mode: the same as twemproxy.
- cluster mode: proxy for redis cluster. You can use non-cluster redis client access the redis cluster.(Inspired with Corvus)
Usage
Build
bash
cargo build --all --release
Run
bash
./target/release/rcproxy default.toml
Install
bash
sudo cp ./target/release/rcproxy /usr/local/bin/
sudo mkdir /etc/rcproxy
sudo cp default.toml /etc/rcproxy/
sudo mkdir /var/log/rcproxy
sudo cp service/systemd/rcproxy.service /lib/systemd/system/
sudo systemctl enable rcproxy
sudo systemctl start rcproxy
Configuration
```Toml
[log]
level = "libaster=info" # "trace" "debug" "info" "warn" "error"
ansi = true # support ANSI colors
stdout = false # print logs to stdout, not to log files
directory = "/var/log/rcproxy" # log file directory
file_name = "rcproxy.log" # log file name
[metrics]
port = 2110
[[clusters]]
name of the cluster. Each cluster means one front-end port.
name="test-redis-cluster"
listen_addr means the cluster font end serve address.
listen_addr="0.0.0.0:9001"
cachetype only support memcache|redis|rediscluster
cachetype="rediscluster"
servers means cache backend. support two format:
for cache_type is memcache or redis, you can set it as:
#
servers = [
"127.0.0.1:7001:10 redis-1",
"127.0.0.1:7002:10 redis-2",
"127.0.0.1:7003:10 redis-3"]
#
as you can see, the format is consisted with:
#
"${addr}:hashweight ${nodealias}"
#
And, for redis_cluster you can set the item as:
#
servers = ["127.0.0.1:7000", "127.0.0.1:7001"]
#
which means the seed nodes to connect to redis cluster.
servers = ["127.0.0.1:7000", "127.0.0.1:7001", "127.0.0.1:7002", "127.0.0.1:7003", "127.0.0.1:7004", "127.0.0.1:7005"]
Work thread number, it's suggested as the number of your cpu(hyper-thread) number.
thread = 1
ReadTimeout is the socket read timeout which effects all in the socket in millisecond
read_timeout = 2000
WriteTimeout is the socket write timeout which effects all in the socket in millisecond
write_timeout = 2000
####################### Cluster Mode Special
fetch means fetch interval for backend cluster to keep cluster info become newer.
default 10 * 60 seconds
fetch = 600
readfromslave is the feature make slave balanced readed by client and ignore side effects.
readfromslave = true
####################### Proxy Mode Special
pingfaillimit means when ping fail reach the limit number, the node will be ejected from the cluster
until the ping is ok in future.
if pingfalilimit == 0, means that close the ping eject feature.
pingfaillimit=3
ping_interval means the interval of each ping was send into backend node in millisecond.
ping_interval=10000
auth = "" # mypassw
```
changelog
see CHANGELOG.md