项目背景:
链路流程: vip —>nginx—>upstream—>server—>ip+port
1.目前还没做路由功能,分组是按服务器分组(不是按用户分组),分组发布过程一个队列形式,这里提供自定义规则切换服upstrem。
2.不管是全量发布,分组发布,快速发布,lb只做一件事:切换upstream(running_upstream与standy_upstream
3.动态切换LB,做法两种,一是直接操作upstream,二是操作upstream里的server.
4.与ci集成方式:超链接web
方案一:模块fly操作upstream,web方式增删查改upsterem里的server
无drop重启nginx
kill -HUP `cat /data/nginx/logs/nginx.pid`
#!/bin/sh
BASE_DIR='/usr/local/'
${BASE_DIR}nginx/sbin/nginx -t -c ${BASE_DIR}nginx/conf/nginx.
conf >& ${BASE_DIR}nginx/logs/nginx.start
conf >& ${BASE_DIR}nginx/logs/nginx.start
info=`cat ${BASE_DIR}nginx/logs/nginx.start`
if [ `echo $info | grep -c "syntax is ok" ` -eq 1 ]; then
if [ `ps aux|grep "nginx"|grep -c "master"` == 1 ]; then
kill -HUP `cat ${BASE_DIR}nginx/logs/nginx.pid`
echo "ok"
else
killall -9 nginx
sleep 1
${BASE_DIR}nginx/sbin/nginx
fi
else
echo "######## error: ########"
cat ${BASE_DIR}nginx/logs/nginx.start
fi
reference:
方案二:模块ngx_dynamic_upstream,api+python方式(直接操作upstream里的server)
因为api接口操作不是太直观,在django上以web方式增删查改upstream
HTTP APIs
You can operate upstreams dynamically with HTTP APIs.
list
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends" server 127.0.0.1:6001; server 127.0.0.1:6002; server 127.0.0.1:6003; $
verbose
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&verbose=" server 127.0.0.1:6001 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6002 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6003 weight=1 max_fails=1 fail_timeout=10; $
update_parameters
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&server=127.0.0.1:6003&weight=10&max_fails=5&fail_timeout=5" server 127.0.0.1:6001 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6002 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6003 weight=10 max_fails=5 fail_timeout=5; $
The supported parameters are blow.
- weight
- max_fails
- fail_timeout
down
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&server=127.0.0.1:6003&down=" server 127.0.0.1:6001 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6002 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6003 weight=1 max_fails=1 fail_timeout=10 down; $
up
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&server=127.0.0.1:6003&up=" server 127.0.0.1:6001 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6002 weight=1 max_fails=1 fail_timeout=10; server 127.0.0.1:6003 weight=1 max_fails=1 fail_timeout=10; $
add
$ curl "http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&add=&server=127.0.0.1:6004" server 127.0.0.1:6001; server 127.0.0.1:6002; server 127.0.0.1:6003; server 127.0.0.1:6004; $
remove
$ curl " http://127.0.0.1:6000/dynamic?upstream=zone_for_backends&remove=&server=127.0.0.1:6003"
server 127.0.0.1:6001;
server 127.0.0.1:6002;
server 127.0.0.1:6004;
server 127.0.0.1:6001;
server 127.0.0.1:6002;
server 127.0.0.1:6004;
reference:
https://github.com/cubicdaiya/ngx_dynamic_upstream
方案三:lua(lua_code_cache)热装载+upstream ,直接操作upstream
简说热装载(热加载) :lua代码产生变更,利用lua_code_cache on ,实现自动加载,不用reload nginx。
亮点: 通过lua处理,不用重启nginx,利用api进行增删查改
不足:操作还是基于命令方式,后期还得加上web操作
restful interface
GET
-
/detail
get all upstreams and their servers -
/list
get the list of upstreams -
/upstream/name
find the upstream by it's name
POST
-
/upstream/name
update one upstream -
body
commands; -
body
server ip:port;
DELETE
-
/upstream/name
delete one upstream
Call the interface, when you get the return code is HTTP_INTERNAL_SERVER_ERROR 500
, you need to reload nginx to make the Nginx work at a good state.
If you got HTTP_CONFLICT 409
, you need resend the same commands again latter.
The /list and /detail interface will return HTTP_NO_CONTENT 204
when there is no upstream.
Other code means you should modify your commands and call the interface again.
ATTENEION
: You also need a third-party
to generate the new config and dump it to Nginx'conf directory.
Sample
» curl -H "host: dyhost" 127.0.0.1:8080502 Bad Gateway 502 Bad Gateway
nginx/1.3.13 » curl -d "server 127.0.0.1:8089;server 127.0.0.1:8088;" 127.0.0.1:8081/upstream/dyhost success » curl -H "host: dyhost" 127.0.0.1:8080 8089 » curl -H "host: dyhost" 127.0.0.1:8080 8088 » curl 127.0.0.1:8081/detail host1 server 127.0.0.1:8088 host2 server 127.0.0.1:8089 dyhost server 127.0.0.1:8089 server 127.0.0.1:8088 » curl -i -X DELETE 127.0.0.1:8081/upstream/dyhost success » curl 127.0.0.1:8081/detail host1 server 127.0.0.1:8088 host2 server 127.0.0.1:8089
API
extern ngx_flag_t ngx_http_dyups_api_enable; ngx_int_t ngx_dyups_update_upstream(ngx_str_t *name, ngx_buf_t *buf, ngx_str_t *rv); ngx_int_t ngx_dyups_delete_upstream(ngx_str_t *name, ngx_str_t *rv); extern ngx_dyups_add_upstream_filter_pt ngx_dyups_add_upstream_top_filter; extern ngx_dyups_del_upstream_filter_pt ngx_dyups_del_upstream_top_filter;
Lua API Example
NOTICE: you should add the directive dyups_interface
into your config file to active this feature
content_by_lua '
local dyups = require "ngx.dyups"
local status, rv = dyups.update("test", [[server 127.0.0.1:8088;]]);
ngx.print(status, rv)
if status ~= ngx.HTTP_OK then
ngx.print(status, rv)
return
end
ngx.print("update success")
status, rv = dyups.delete("test")
if status ~= ngx.HTTP_OK then
ngx.print(status, rv)
return
end
ngx.print("delete success")
local dyups = require "ngx.dyups"
local status, rv = dyups.update("test", [[server 127.0.0.1:8088;]]);
ngx.print(status, rv)
if status ~= ngx.HTTP_OK then
ngx.print(status, rv)
return
end
ngx.print("update success")
status, rv = dyups.delete("test")
if status ~= ngx.HTTP_OK then
ngx.print(status, rv)
return
end
ngx.print("delete success")
';
reference:
https://github.com/yzprofile/ngx_http_dyups_module
https://github.com/moonbingbing/openresty-best-practices/blob/master/ngx_lua/hot_load.md
方案四:自研
自已写lua+redis,再封装一层api,供业务插入标签,再进行增删查改