Lua resty.http模块问题？

eagle-china

使用resty.http发送http请求，设置了keepalive后。怎么我刷新一次就会有新的socket被建立，是不是我的keepalive没有生效呢？还是我使用的方法不对？

location / {

root /var/wwwroot/;

content_by_lua_file '/lua/test_proxy.lua';

index index.html index.htm;

}

test_proxy.lua 代码如下:

local http = require 'resty.http'

local httpc = http.new()

httpc:set_timeout(5000)

local ok, err = httpc:connect('12.104.51.10', 80)

if not ok then

ngx.log(ngx.ERR, err)

return

end

httpc:set_timeout(5000)

httpc:proxy_response(httpc:proxy_request(), 8192)

httpc:set_keepalive(90000, 3)

以下是每刷新一次的socket连接情况(每次刷新的时间间隔在keepalive的max_idle_timeout参数之内):

[sy@Node-02:~]

% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED

(No info could be read for "-p": geteuid()=1000 but you should be root.)

tcp 0 0 192.168.56.131:1150 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -

[sy@Node-02:~]

% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED

(No info could be read for "-p": geteuid()=1000 but you should be root.)

tcp 0 0 192.168.56.131:1151 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -

[sy@Node-02:~]

% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED

(No info could be read for "-p": geteuid()=1000 but you should be root.)

tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -

[sy@Node-02:~]

% netstat -tanlp|grep 12.104.51.10

(No info could be read for "-p": geteuid()=1000 but you should be root.)

tcp 0 0 192.168.56.131:1150 12.104.51.10:80 TIME_WAIT -

tcp 0 0 192.168.56.131:1148 12.104.51.10:80 TIME_WAIT -

tcp 0 0 192.168.56.131:1151 12.104.51.10:80 TIME_WAIT -

tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -

tcp 0 0 192.168.56.131:1144 12.104.51.10:80 TIME_WAIT -

tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -

yuansheng_2f16

tcpsock:setkeepalive(timeout?, size?)

https://www.nginx.com/resources/wiki/modules/lua/#tcpsock-setkeepalive

看你的示例代码：httpc:set_keepalive(90000, 3)

你这里只设置了3个长连接，这是你期望的？

另外：开启长连接时，lua_code_cache 必须要设置为 on 。

希望对您有用。

2015-09-29 16:34 GMT+08:00 X山鹰 <eagle...@hotmail.com>:

使用resty.http发送http请求，设置了keepalive后。怎么我刷新一次就会有新的socket被建立，是不是我的keepalive没有生效呢？还是我使用的方法不对？

location / {
root /var/wwwroot/;
content_by_lua_file '/lua/test_proxy.lua';
index index.html index.htm;
}

test_proxy.lua 代码如下:

local http = require 'resty.http'

local httpc = http.new()
httpc:set_timeout(5000)
local ok, err = httpc:connect('12.104.51.10', 80)
if not ok then
ngx.log(ngx.ERR, err)
return
end

httpc:set_timeout(5000)
httpc:proxy_response(httpc:proxy_request(), 8192)
httpc:set_keepalive(90000, 3)

以下是每刷新一次的socket连接情况(每次刷新的时间间隔在keepalive的max_idle_timeout参数之内):
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1148 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -

--

Membhis

My github: https://github.com/membphis

Our Book: OpenResty Best Practices

eagle-china

我的连接池大小为3、lua_code_cache 也设置了on。现在的问题是每次请求都会重新建立socket，之前的socket会TIME_WAIT。你可以看我下面用netstat查看socket的情况。

From: yuan...@iresty.org
Date: Tue, 29 Sep 2015 19:27:29 +0800
Subject: Re: [openresty] Lua resty.http模块问题？
To: openresty@googlegroups.com

tcpsock:setkeepalive(timeout?, size?)

https://www.nginx.com/resources/wiki/modules/lua/#tcpsock-setkeepalive

看你的示例代码：httpc:set_keepalive(90000, 3)

你这里只设置了3个长连接，这是你期望的？

另外：开启长连接时，lua_code_cache 必须要设置为 on 。

希望对您有用。

2015-09-29 16:34 GMT+08:00 X山鹰 <eagle...@hotmail.com>:

使用resty.http发送http请求，设置了keepalive后。怎么我刷新一次就会有新的socket被建立，是不是我的keepalive没有生效呢？还是我使用的方法不对？

location / {
root /var/wwwroot/;
content_by_lua_file '/lua/test_proxy.lua';
index index.html index.htm;
}

test_proxy.lua 代码如下:

local http = require 'resty.http'

local httpc = http.new()
httpc:set_timeout(5000)
local ok, err = httpc:connect('12.104.51.10', 80)
if not ok then
ngx.log(ngx.ERR, err)
return
end

httpc:set_timeout(5000)
httpc:proxy_response(httpc:proxy_request(), 8192)
httpc:set_keepalive(90000, 3)

以下是每刷新一次的socket连接情况(每次刷新的时间间隔在keepalive的max_idle_timeout参数之内):
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1148 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -

--

Membhis

My github: https://github.com/membphis

Our Book: OpenResty Best Practices

doujiang24

Hello,

还有一种可能情况是，你配置了多个 nginx worker，而连接池是基于每 worker 的

在 2015年9月30日上午9:20，X山鹰 <eagle...@hotmail.com>写道：

我的连接池大小为3、lua_code_cache 也设置了on。现在的问题是每次请求都会重新建立socket，之前的socket会TIME_WAIT。你可以看我下面用netstat查看socket的情况。

From: yuan...@iresty.org
Date: Tue, 29 Sep 2015 19:27:29 +0800
Subject: Re: [openresty] Lua resty.http模块问题？
To: openresty@googlegroups.com

tcpsock:setkeepalive(timeout?, size?)
https://www.nginx.com/resources/wiki/modules/lua/#tcpsock-setkeepalive

看你的示例代码：httpc:set_keepalive(90000, 3)

你这里只设置了3个长连接，这是你期望的？

另外：开启长连接时，lua_code_cache 必须要设置为 on 。

希望对您有用。

2015-09-29 16:34 GMT+08:00 X山鹰 <eagle...@hotmail.com>:

使用resty.http发送http请求，设置了keepalive后。怎么我刷新一次就会有新的socket被建立，是不是我的keepalive没有生效呢？还是我使用的方法不对？

location / {
root /var/wwwroot/;
content_by_lua_file '/lua/test_proxy.lua';
index index.html index.htm;
}

test_proxy.lua 代码如下:

local http = require 'resty.http'

local httpc = http.new()
httpc:set_timeout(5000)
local ok, err = httpc:connect('12.104.51.10', 80)
if not ok then
ngx.log(ngx.ERR, err)
return
end

httpc:set_timeout(5000)
httpc:proxy_response(httpc:proxy_request(), 8192)
httpc:set_keepalive(90000, 3)

以下是每刷新一次的socket连接情况(每次刷新的时间间隔在keepalive的max_idle_timeout参数之内):
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10|grep ESTABLISHED
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -
[sy@Node-02:~]
% netstat -tanlp|grep 12.104.51.10
(No info could be read for "-p": geteuid()=1000 but you should be root.)
tcp 0 0 192.168.56.131:1150 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1148 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1151 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1149 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1152 12.104.51.10:80 ESTABLISHED -
tcp 0 0 192.168.56.131:1144 12.104.51.10:80 TIME_WAIT -
tcp 0 0 192.168.56.131:1153 12.104.51.10:80 ESTABLISHED -

--

--

Membhis
My github: https://github.com/membphis
Our Book: OpenResty Best Practices

--

--

agentzh

Hello!

2015-09-29 16:34 GMT+08:00 X山鹰:
> httpc:set_timeout(5000)
> httpc:proxy_response(httpc:proxy_request(), 8192)
> httpc:set_keepalive(90000, 3)
>

如德江所言，连接池是每 worker 的，所以如果请求撞上另一个还没有建立连接池的 worker，则会创建新的连接，因为无空闲连接可复用。

另外，你应当总是对你这里的各个可能会返回错误的函数调用进行错误处理。比如 set_keepalive 可能会失败。这样你可以有机会得到具体的错误信息。

Regards,
-agentzh

eagle-china

谢谢各位的帮助^_^

另外resty.http 模块对post请求的支持是不是不太好呀？我发送post请求一直不成功呢。后端server一直没有收到post的数据。

> Date: Wed, 30 Sep 2015 10:38:37 +0800
> Subject: Re: [openresty] Lua resty.http模块问题？
> From: age...@gmail.com
> To: openresty@googlegroups.com
>
> Hello!
>
> 2015-09-29 16:34 GMT+08:00 X山鹰:
> > httpc:set_timeout(5000)
> > httpc:proxy_response(httpc:proxy_request(), 8192)
> > httpc:set_keepalive(90000, 3)
> >
>
> 如德江所言，连接池是每 worker 的，所以如果请求撞上另一个还没有建立连接池的 worker，则会创建新的连接，因为无空闲连接可复用。
>
> 另外，你应当总是对你这里的各个可能会返回错误的函数调用进行错误处理。比如 set_keepalive 可能会失败。这样你可以有机会得到具体的错误信息。
>
> Regards,
> -agentzh
>
> --
> --
> 邮件来自列表“openresty”,专用于技术讨论!
> 订阅: 请发空白邮件到 openresty+subscribe@googlegroups.com
> 发言: 请发邮件到 openresty@googlegroups.com
> 退订: 请发邮件至 openresty+unsubscribe@googlegroups.com
> 归档: http://groups.google.com/group/openresty
> 官网: http://openresty.org/
> 仓库: https://github.com/agentzh/ngx_openresty
> 教程: http://openresty.org/download/agentzh-nginx-tutorials-zhcn.html

agentzh

Hello!

2015-09-30 16:36 GMT+08:00 X山鹰:
> 另外resty.http 模块对post请求的支持是不是不太好呀？ 我发送post请求一直不成功呢 。后端server一直没有收到post的数据。
>

你忘记设置 Content-Type 请求头了吧？

调试这种问题，使用 nc, wireshark 或者 tcpdump 这样的工具在线路上进行检查是很方便的。

Regards,
-agentzh

eagle-china

我用tcpdump抓包请求头都正常也包括Content-Type 只是没有post的数据而已？

0x0000: 4500 0238 0da7 4000 4006 2179 c0a8 3883 E..8..@.@.!y..8.

0x0010: 790e 9766 2bd2 0050 c4b5 346f f13c 03b4 y..f+..P..4o.<..

0x0020: 5018 3908 0bcb 0000 504f 5354 202f 636f P.9.....POST./co

0x0030: 6e74 6163 7420 4854 5450 2f31 2e31 0d0a ntact.HTTP/1.1..

0x0040: 6163 6365 7074 2d6c 616e 6775 6167 653a accept-language:

0x0050: 207a 682d 434e 2c7a 683b 713d 302e 380d .zh-CN,zh;q=0.8.

0x0060: 0a63 6f6e 7465 6e74 2d74 7970 653a 2061 .content-type:.a

0x0070: 7070 6c69 6361 7469 6f6e 2f78 2d77 7777 pplication/x-www

0x0080: 2d66 6f72 6d2d 7572 6c65 6e63 6f64 6564 -form-urlencoded

0x0090: 3b20 6368 6172 7365 743d 5554 462d 380d ;.charset=UTF-8.

0x00a0: 0a63 6f6e 6e65 6374 696f 6e3a 206b 6565 .connection:.kee

0x00b0: 702d 616c 6976 650d 0a63 6f6e 7465 6e74 p-alive..content

0x00c0: 2d6c 656e 6774 683a 2038 320d 0a63 6f6f -length:.82..coo

0x00d0: 6b69 653a 2061 6363 6573 7365 643d 7472 kie:.accessed=tr

0x00e0: 7565 0d0a 6163 6365 7074 2d65 6e63 6f64 ue..accept-encod

0x00f0: 696e 673a 2067 7a69 702c 2064 6566 6c61 ing:.gzip,.defla

0x0100: 7465 0d0a 686f 7374 3a20 7777 772e 6c69 te..host:.www.li

0x0110: 742e 6d6e 0d0a 6163 6365 7074 3a20 2a2f t.mn..accept:.*/

0x0120: 2a0d 0a78 2d72 6571 7565 7374 6564 2d77 *..x-requested-w

0x0130: 6974 683a 2058 4d4c 4874 7470 5265 7175 ith:.XMLHttpRequ

0x0140: 6573 740d 0a6f 7269 6769 6e3a 2068 7474 est..origin:.htt

0x0150: 703a 2f2f 7777 772e 6c69 742e 6d6e 0d0a p://12.104.51.10..

0x0160: 7072 6167 6d61 3a20 6e6f 2d63 6163 6865 pragma:.no-cache

0x0170: 0d0a 7265 6665 7265 723a 2068 7474 703a ..referer:.http:

0x0180: 2f2f 7777 772e 6c69 742e 6d6e 2f63 6f6e //12.104.51.10/con

0x0190: 7461 6374 0d0a 6361 6368 652d 636f 6e74 tact..cache-cont

0x01a0: 726f 6c3a 206e 6f2d 6361 6368 650d 0a75 rol:.no-cache..u

0x01b0: 7365 722d 6167 656e 743a 204d 6f7a 696c ser-agent:.Mozil

0x01c0: 6c61 2f35 2e30 2028 4d61 6369 6e74 6f73 la/5.0.(Macintos

0x01d0: 683b 2049 6e74 656c 204d 6163 204f 5320 h;.Intel.Mac.OS.

0x01e0: 5820 3130 5f31 305f 3529 2041 7070 6c65 X.10_10_5).Apple

0x01f0: 5765 624b 6974 2f35 3337 2e33 3620 284b WebKit/537.36.(K

0x0200: 4854 4d4c 2c20 6c69 6b65 2047 6563 6b6f HTML,.like.Gecko

0x0210: 2920 4368 726f 6d65 2f34 342e 302e 3234 ).Chrome/44.0.24

0x0220: 3033 2e31 3537 2053 6166 6172 692f 3533 03.157.Safari/53

0x0230: 372e 3336 0d0a 0d0a 7.36....

16:32:59.214184 IP 12.104.51.10.80 > 192.168.56.131.11218: Flags [.], ack 529, win 64240, length 0

0x0000: 4500 0028 21e2 0000 8006 0f4e 790e 9766 E..(!......Ny..f

0x0010: c0a8 3883 0050 2bd2 f13c 03b4 c4b5 367f ..8..P+..<....6.

0x0020: 5010 faf0 8efb 0000 0000 0000 0000 P.............

16:33:59.332166 IP 12.104.51.10.80 > 192.168.56.131.11218: Flags [FP.], seq 1, ack 529, win 64240, length 0

0x0000: 4500 0028 21f1 0000 8006 0f3f 790e 9766 E..(!......?y..f

0x0010: c0a8 3883 0050 2bd2 f13c 03b4 c4b5 367f ..8..P+..<....6.

0x0020: 5019 faf0 8ef2 0000 0000 0000 0000 P.............

16:33:59.332298 IP 192.168.56.131.11218 > 12.104.51.10.80: Flags [F.], seq 529, ack 2, win 14600, length 0

0x0000: 4500 0028 0da8 4000 4006 2388 c0a8 3883 E..(..@.@.#...8.

0x0010: 790e 9766 2bd2 0050 c4b5 367f f13c 03b5 y..f+..P..6..<..

0x0020: 5011 3908 09bb 0000 P.9.....

16:33:59.332558 IP 12.104.51.10.80 > 192.168.56.131.11218: Flags [.], ack 530, win 64239, length 0

0x0000: 4500 0028 21f2 0000 8006 0f3e 790e 9766 E..(!......>y..f

0x0010: c0a8 3883 0050 2bd2 f13c 03b5 c4b5 3680 ..8..P+..<....6.

0x0020: 5010 faef 8efa 0000 0000 0000 0000 P.............

> Date: Wed, 30 Sep 2015 16:39:02 +0800
> Subject: Re: [openresty] Lua resty.http模块问题？
> From: age...@gmail.com
> To: openresty@googlegroups.com
>
> Hello!
>
> 2015-09-30 16:36 GMT+08:00 X山鹰:
> > 另外resty.http 模块对post请求的支持是不是不太好呀？我发送post请求一直不成功呢。后端server一直没有收到post的数据。
> >
>
> 你忘记设置 Content-Type 请求头了吧？
>
> 调试这种问题，使用 nc, wireshark 或者 tcpdump 这样的工具在线路上进行检查是很方便的。
>
> Regards,
> -agentzh
>
> --
> --
> 邮件来自列表“openresty”,专用于技术讨论!
> 订阅: 请发空白邮件到 openresty+subscribe@googlegroups.com
> 发言: 请发邮件到 openresty@googlegroups.com
> 退订: 请发邮件至 openresty+unsubscribe@googlegroups.com
> 归档: http://groups.google.com/group/openresty
> 官网: http://openresty.org/
> 仓库: https://github.com/agentzh/ngx_openresty
> 教程: http://openresty.org/download/agentzh-nginx-tutorials-zhcn.html

eagle-china

我看了一下resty.http的源码, 发现proxy_request底层是调用send_request方法。我在该方法中输出body发现是空得呢

function _M.send_request(self, params)

-- Apply defaults

setmetatable(params, { __index = DEFAULT_PARAMS })

local sock = self.sock

local body = params.body

local headers = http_headers.new()

local params_headers = params.headers

if params_headers then

-- We assign one by one so that the metatable can handle case insensitivity

-- for us. You can blame the spec for this inefficiency.

for k,v in pairs(params_headers) do

headers[k] = v

end

// ...... 此处省略

agentzh

Hello!

2015-09-30 17:03 GMT+08:00 X山鹰:
> 我看了一下resty.http的源码, 发现proxy_request底层是调用send_request方法。我在该方法中输出body发现是空得呢
>
> function _M.send_request(self, params)
>     -- Apply defaults
>     setmetatable(params, { __index = DEFAULT_PARAMS })
>
>     local sock = self.sock
>     local body = params.body

那么显然你自己没有把请求体数据传递给 params.body?

Regards,
-agentzh

eagle-china

function _M.proxy_request(self, chunksize)

return self:request{

method = ngx_req_get_method(),

path = ngx_var.uri .. ngx_var.is_args .. (ngx_var.query_string or ""),

body = self:get_client_body_reader(chunksize),

headers = ngx_req_get_headers(),

}

end

嗯，我把proxy_request方法中body的值从 self:get_client_body_reader(chunksize) 替换ngx.encode_args(ngx.req.get_post_args())发送POST/GET请求就正常。

self:get_client_body_reader(chunksize)和ngx.req.get_post_args()有什么不一样吗？为什么前者再发送POST时没有body数据呢？

> Date: Thu, 1 Oct 2015 10:36:40 +0800
> Subject: Re: [openresty] Lua resty.http模块问题？
> From: age...@gmail.com
> To: openresty@googlegroups.com
>
> Hello!
>
> 2015-09-30 17:03 GMT+08:00 X山鹰:
> > 我看了一下resty.http的源码, 发现proxy_request底层是调用send_request方法。我在该方法中输出body发现是空得呢
> >
> > function _M.send_request(self, params)
> > -- Apply defaults
> > setmetatable(params, { __index = DEFAULT_PARAMS })
> >
> > local sock = self.sock
> > local body = params.body
>
> 那么显然你自己没有把请求体数据传递给 params.body?
>
> Regards,
> -agentzh
>
> --
> --
> 邮件来自列表“openresty”,专用于技术讨论!
> 订阅: 请发空白邮件到 openresty+subscribe@googlegroups.com
> 发言: 请发邮件到 openresty@googlegroups.com
> 退订: 请发邮件至 openresty+unsubscribe@googlegroups.com
> 归档: http://groups.google.com/group/openresty
> 官网: http://openresty.org/
> 仓库: https://github.com/agentzh/ngx_openresty
> 教程: http://openresty.org/download/agentzh-nginx-tutorials-zhcn.html