savsgio/kratgo 基于fasthttp 开发的web cache 服务，包含以下特性
支持的特性
- cache proxy
- lb （顺序，然后轮转模式，没有权重）
- 配置cache 规则
- 配置http 请求头
- 管理api 支持，基于管理api 我们可以进行cache 清理
参考使用
- docker-compose 文件
version: "3"
services:
proxy:
image: savsgio/kratgo
command: -config /etc/config.yaml
ports:
- "6081:6081"
volumes:
- "./config.yaml:/etc/config.yaml"
- 配置
###########################
# KRATGO CONFIGURATION
###########################
# IMPORTANT: Be careful with the tabulation indentation
# --- Variables ---
# $(method) : request method
# $(host) : request host
# $(path) : request path
# $(contentType) : response backend's content type
# $(statusCode) : response backend's status code
# $(req.header::<NAME>) : request header name
# $(resp.header::<NAME>) : response header name
# $(cookie::<NAME>) : request cookie name
# --- Operators ---
# Modifiers: + - / * & | ^ ** % >> <<
# Comparators: > >= < <= == != =~ !~
# Logical ops: || &&
# Numeric constants, as 64-bit floating point (12345.678)
# String constants (single quotes: 'foobar')
# Date constants (single quotes, using any permutation of RFC3339, ISO8601, ruby date, or unix date; date parsing is automatically tried with any string constant)
# Boolean constants: true false
# Parenthesis to control order of evaluation ( )
# Arrays (anything separated by , within parenthesis: (1, 2, 'foo'))
# Prefixes: ! - ~
# Ternary conditional: ? :
# Null coalescence: ??
# --- Log ---
# Log level: fatal | error | warning | info | debug
# Log output:
# - console: Write output in standard error
# - <file path>: Write output in log file
logLevel: debug
logOutput: ./kratgo.log
# --- Cache ---
# ttl: Cache expiration in minutes
# cleanFrequency: Interval in minutes between removing expired entries (clean up)
# maxEntries: Max number of entries in cache. Used only to calculate initial size for cache
# maxEntrySize: Max size of entry in bytes
# hardMaxCacheSize: Limit for cache size in MB (Default value is 0 which means unlimited size)
cache:
ttl: 10
cleanFrequency: 1
maxEntries: 600000
maxEntrySize: 500
hardMaxCacheSize: 0
# --- Invalidator ---
# maxWorkers: Maximum workers to execute invalidations
invalidator:
maxWorkers: 5
# --- Proxy ---
# addr: IP and Port of Kratgo
# backendAddrs: Array with "addr:port" of the backends
# response: Configuration to manipulate reponse (Optional)
# headers:
# set: Configuration to SET headers from response (Optional)
# - name: Header name
# value: Value of header
# if: Condition to set this header (Optional)
#
# unset: Configuration to UNSET headers from response (Optional)
# - name: Header name
# if: Condition to unset this header (Optional)
#
# nocache: Conditions to not save in cache the backend response (Optional)
proxy:
addr: 0.0.0.0:6081
backendAddrs:
[
<address>
]
response:
headers:
set:
- name: X-Kratgo
value: true
unset:
- name: Set-Cookie
if: $(req.header::X-Requested-With) != 'XMLHttpRequest'
nocache:
- $(req.header::X-Requested-With) == 'XMLHttpRequest'
# --- Admin ---
# addr: IP and Port of admin api
admin:
addr: 0.0.0.0:6082
- 效果
同时，如果断开网络链接，你会发现网站依然还能访问（因为cache了） 
基于fasthttp 开发一个自己proxy 的简单代码
package main
import (
"flag"
"fmt"
"log"
"github.com/valyala/fasthttp"
)
var (
add = flag.String("add", "localhost", "proxy address")
)
type myLog struct {
add string
}
func (l myLog) Printf(format string, args
log.Printf(format, args
}
func (l myLog) myhandler(ctx *fasthttp.RequestCtx) {
backend := fasthttp.HostClient{
Addr: l.add,
}
backend.Do(&ctx.Request, &ctx.Response)
}
func main() {
flag.Parse()
fmt.Println(*add)
log := myLog{
add: *add,
}
s := fasthttp.Server{
Handler: log.myhandler,
LogAllErrors: true,
Logger: log,
}
s.ListenAndServe(":8000")
}
源码简单说明
savsgio/kratgo 的源码并不难，对于cache proxy 基于fasthttp 的Server提供，对于cache 部分基于bigcache
Server handler 使用了fasthttp 的fasthttp.HostClient,配置解析部分基于yaml，指定语法解析（配置自定义参数）
使用Knetic/govaluate，同时为了性能作者很多地方使用了sync.Pool ,lb 部分实现很简单，就是数据索引的移动
当超过数据最大值得时候，重置索引为0，同时对于cache 管理，作者基于channel 实现了一个校验任务，实际可以
看看源码，实际上以后也分享过gregjones/httpcache的开源实现，也是一个不错的选择
参考资料
https://github.com/savsgio/kratgo
https://github.com/Knetic/govaluate
https://github.com/valyala/fasthttp
https://github.com/allegro/bigcache
https://github.com/gregjones/httpcache