guycheung / falcon-ngx_metric Goto Github PK

python nginx_collect.py --service=hostname --format=falcon --falcon-addr=http://127.0.0.1:1988/v1/push
Traceback (most recent call last):
File "nginx_collect.py", line 451, in collect
append_datapoint(datapoints, Render.render(la))
File "nginx_collect.py", line 71, in render
c = renders[la[0]]
KeyError: ''

这个报错是 采集参数拿到一个html作为 key ?

只设置这个参数无效，有人改过这个参数吗？

客户端打这种请求： { "timestamp": "2018-04-18T01:02:34+08:00", "host": "xx.xx.xx.xxx", "user": "-", "request": "\x16\x03\x01\x00\xDA\x01\x00\x00\xD6\x03\x03\xAD\x03\x99&*{\xAB\xAB\x8C\x1DS\x82\xCF\xA4y\xCF\x84\xB4\xEB\xAF}\x93S,u\xD4\xB5\x1E\x10\xB2\xE5;\x00\x00(\xC0,\xC0+\xC0$\xC0#\xC0", "clientip": "xx.xxx.xx.xxx", "size": 166, "responsetime": 5.000, "upstreamtime": "-", "upstreamhost": "-", "upstreamstatus": "-", "http_host": "_", "url": "-", "referrer": "-", "x_forwarded_for": "-", "agent": "-", "status": "400"}

error日志会有这种错误
2018/04/18 10:02:30 [error] 124908#124908: 4252054 failed to run log_by_lua: /usr/local/nginx/modules/ngx_metric/util.lua:49: bad argument #1 to 'gmatch' (string expected, got nil)
stack traceback:
[C]: in function 'gmatch'
/usr/local/nginx/modules/ngx_metric/util.lua:49: in function 'str_split'
.../local/nginx/modules/ngx_metric/ngx_metric.lua:12: in function 'cut_uri'
.../local/nginx/modules/ngx_metric/ngx_metric.lua:24: in function 'new'
/usr/local/nginx/modules/ngx_metric.lua:16: in function </usr/local/nginx/modules/ngx_metric.lua:1> while logging request, client: xxx.xx.xx.xxx, server: _, request: "µ±Z֧68Ӝe^,>c5T~쀍"

Traceback (most recent call last):
File "nginx_collect.py", line 450, in collect
append_datapoint(datapoints, Render.render(la))
File "nginx_collect.py", line 70, in render
c = renders[la[0]]
KeyError: ''

in falcon-ngx_metric/lua/ngx_metric/histogram.lua 24 line
if we should use **self.**values = util.str_split(values, ",") instead?

[root@izuf68l0q1ggeg7q97i1o5z falcon-ngx_metric]# systemctl restart nginx
Job for nginx.service failed because the control process exited with error code. See "systemctl status nginx.service" and "journalctl -xe" for details.
[root@izuf68l0q1ggeg7q97i1o5z falcon-ngx_metric]# systemctl status nginx.service
Jan 13 15:16:27 izuf68l0q1ggeg7q97i1o5z nginx[2388]: nginx: [emerg] unknown directive "lua_package_path" in /etc/nginx/conf.d/ngx_metric.conf:1

在dashboard中经常出现断点，重新恢复的请求量差不多是“失联”这段时间的请求量，请问该如何排查，机器openfalcon排查没问题，agent也没问题

2016/12/08 14:46:13 [error] 25707#0: *633 lua entry thread aborted: runtime error: modules/ngx_metric/util.lua:9: attempt to call method 'get_keys' (a nil value)
stack traceback:
coroutine 0:
modules/ngx_metric/util.lua: in function 'dict_get_keys'
modules/ngx_metric/counter.lua:18: in function 'get_snapshot'
/usr/local/nginx/modules/ngx_metric_output.lua:18: in function </usr/local/nginx/modules/ngx_metric_output.lua:1>, client: 127.0.0.1, server: 127.0.0.1, request: "GET /monitor/basic_status HTTP/1.1", host: "localhost:9091"
2016/12/08 14:46:13 [error] 25707#0: 633 failed to run log_by_lua: modules/ngx_metric/util.lua:35: attempt to call method 'safe_add' (a nil value)
stack traceback:
modules/ngx_metric/util.lua:35: in function 'dict_safe_incr'
modules/ngx_metric/counter.lua:14: in function 'add'
modules/ngx_metric/ngx_metric.lua:58: in function 'err_count'
modules/ngx_metric/ngx_metric.lua:92: in function 'record'
/usr/local/nginx/modules/ngx_metric.lua:17: in function </usr/local/nginx/modules/ngx_metric.lua:1> while logging request, client: 127.0.0.1, server: 127.0.0.1, request: "GET /monitor/basic_status HTTP/1.1", host: "localhost:9091"

咨询下这个status_code 状态码的定义是写死的吗？比如我只想要502的数量，这种可以吗？

你好：
鉴于每个人的nginx启动方式和位置不同，可以考虑提示下部分代码内的lua文件路径地址根据需要修改，目前ngx_metric.conf 文件里调用的文件路径是相对的，会报错找不到文件。
别问我怎么想到的。嘿嘿。丢人。

error_rate
搜索失败：no counters in graph

有采集流量这个字段吗？请问latency_{50,75,95}th这几个字段表示什么意思呢？采集的数据大体一致，只是有些点不一样

一个负载均衡下的两个相同的nginx，运行着两个falcon-ngx_metric，相同的api请求数目统计push上去会相加么？

nginx版本：1.8.0
nginx编译参数：

nginx version: nginx/1.8.0
built by gcc 4.8.5 20150623 (Red Hat 4.8.5-11) (GCC) 
built with OpenSSL 1.0.2e 3 Dec 2015
TLS SNI support enabled
configure arguments: --prefix=/usr/local/services/nginx-1.8.0 --with-pcre=/data/tmp/pcre-8.37 --with-zlib=/data/tmp/zlib-1.2.8 --with-openssl=/data/tmp/openssl-1.0.2e --with-http_gzip_static_module --with-http_ssl_module --add-module=/data/tmp/ngx_http_qrcode_module --with-ld-opt=-Wl,-rpath,/usr/local/lib --add-module=/data/tmp/ngx_http_barcode --add-module=/usr/local/src/lua-nginx-module-0.10.10/ --add-module=/usr/local/src/ngx_devel_kit-0.3.0/ --with-http_stub_status_module

lua版本：
ll /usr/local/bin/luajit
lrwxrwxrwx 1 root root 12 Nov 14 10:43 /usr/local/bin/luajit -> luajit-2.0.5

nginx_metric配置：

lua_package_path "modules/?.lua;;";
lua_shared_dict result_dict 128M;
log_by_lua_file modules/ngx_metric.lua;

server {

    listen          127.0.0.1:9091;
    server_name     127.0.0.1;
    access_log logs/ngx_metric_for_falcon.access.log;
    error_log logs/ngx_metric_for_falcon.error.log;

    location /monitor/basic_status {
        content_by_lua_file modules/ngx_metric_output.lua;
        access_log off;
        allow 127.0.0.1;
        deny all;
    }   

    location /monitor/nginx_status {
        stub_status on; 
        access_log off ;
        allow 127.0.0.1;
        deny all;
    }   
}

配置好nginx，请求：
curl http://127.0.0.1:9091/monitor/basic_status
返回500错误

nginx_metric.error.log:

2018/11/14 17:11:31 [error] 21186#0: *8304 lua entry thread aborted: runtime error: ...local/services/nginx-1.8.0/modules/ngx_metric_output.lua:3: module 'ngx_metric.counter' not found:
	no field package.preload['ngx_metric.counter']
	no file 'modules/ngx_metric/counter.lua'
	no file './ngx_metric/counter.lua'
	no file '/usr/local/share/luajit-2.0.5/ngx_metric/counter.lua'
	no file '/usr/local/share/lua/5.1/ngx_metric/counter.lua'
	no file '/usr/local/share/lua/5.1/ngx_metric/counter/init.lua'
	no file './ngx_metric/counter.so'
	no file '/usr/local/lib/lua/5.1/ngx_metric/counter.so'
	no file '/usr/local/lib/lua/5.1/loadall.so'
	no file './ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/loadall.so'
stack traceback:
coroutine 0:
	[C]: in function 'require'
	...local/services/nginx-1.8.0/modules/ngx_metric_output.lua:3: in function <...local/services/nginx-1.8.0/modules/ngx_metric_output.lua:1>, client: 127.0.0.1, server: 127.0.0.1, request: "GET /monitor/basic_status HTTP/1.1", host: "127.0.0.1:9091"
2018/11/14 17:11:31 [error] 21186#0: *8304 failed to run log_by_lua*: /usr/local/services/nginx-1.8.0/modules/ngx_metric.lua:4: module 'ngx_metric.ngx_metric' not found:
	no field package.preload['ngx_metric.ngx_metric']
	no file 'modules/ngx_metric/ngx_metric.lua'
	no file './ngx_metric/ngx_metric.lua'
	no file '/usr/local/share/luajit-2.0.5/ngx_metric/ngx_metric.lua'
	no file '/usr/local/share/lua/5.1/ngx_metric/ngx_metric.lua'
	no file '/usr/local/share/lua/5.1/ngx_metric/ngx_metric/init.lua'
	no file './ngx_metric/ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/ngx_metric/ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/loadall.so'
	no file './ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/ngx_metric.so'
	no file '/usr/local/lib/lua/5.1/loadall.so'
stack traceback:
	[C]: in function 'require'
	/usr/local/services/nginx-1.8.0/modules/ngx_metric.lua:4: in function </usr/local/services/nginx-1.8.0/modules/ngx_metric.lua:1> while logging request, client: 127.0.0.1, server: 127.0.0.1, request: "GET /monitor/basic_status HTTP/1.1", host: "127.0.0.1:9091"

看了下代码，写得很赞，学习了很多，特别是对sample抽样部分。
有两个问题请教下：

1，目前这个统计是针对uri（location）层面的，我打算将其扩展到可以统计server层和http层的，不知道有啥难度没？（估计应该是没啥难度的，代码的耦合度都很低，很方便扩展）

2，另外，是否可以按照每个upstream的名字统计下每个upstream的counter和time? 这样更具体详细。
(粗略看了下，目前只是按照uri对相应的upstream进行了time的统计,获取upstream的名字会是难点么？）

按照教程配置好了，但是

curl http://127.0.0.1:9091/monitor/basic_status

之后，只能得到query_count 和 lantency信息，如下所示

query_count|127.0.0.1|/monitor/basic_status|3
latency_99th|127.0.0.1|/monitor/basic_status|0.002
latency_95th|127.0.0.1|/monitor/basic_status|0.002
latency_50th|127.0.0.1|/monitor/basic_status|0.001
latency_75th|127.0.0.1|/monitor/basic_status|0.002

请问是什么问题 🙏

Traceback (most recent call last):
File "nginx_collect.py", line 450, in collect
append_datapoint(datapoints, Render.render(la))
File "nginx_collect.py", line 70, in render
c = renders[la[0]]
KeyError: ''

我执行脚本以后就报下面的错，请问是什么导致的呢，谢谢！
[root@fqy-nginx-4 falcon-ngx_metric]# python nginx_collect.py --format=falcon
Traceback (most recent call last):
File "nginx_collect.py", line 450, in collect
append_datapoint(datapoints, Render.render(la))
File "nginx_collect.py", line 70, in render
c = renders[la[0]]
KeyError: ''

/civil/api/civil.user.UserLogin
/civil/api/civil.device.GetDeviceListEx
/civil/api/civil.device.SetAlarmPlanConfig
。。。
upstream_web_civil 这个回源接口的req_total_count、 timeout_ total_count、 error_ total_count、 latency_ total_avg

比如我只想要过滤出如上这样的几个接口，其他都屏蔽掉， 有什么方法么？

/monitor/nginx_status， 您在nginx.conf有配置这个，但我发现代码中并没有通过这个接口来获取连接数等，这个配置是不是可以去掉呢？

大神，请问ngx_metric收集了当前nginx所有的访问吗？ /monitor/basic_status这个链接的请求，是否记录了所有？

执行脚本之后可以取得数据 貌似没有推送到agent端 有没有跟我遇到一样的情况的

使用过程中，我按照 README.md 的方法，部署好了 lua 模块。但遇到些问题，我的使用流程大致如下：

1. 通过如下 bash 脚本来模拟访问 lua 模块监控的站点

#!/bin/bash

while [ 1 == 1 ]; do
	curl http://localhost:9091/ >/dev/null 2>&1
	sleep 1s
	curl http://localhost:9091/monitor/basic_status >/dev/null 2>&1
	curl http://localhost:9091/monitor/nginx_status >/dev/null 2>&1
	curl http://localhost:9091/hello >/dev/null 2>&1
	sleep 1s
done

2. 通过如下 bash 脚本来 push 搜集到的数据到 agent 接口（http://127.0.0.1:1988/v1/push）

#!/bin/bash

while [ 1 == 1 ]; do
        python nginx_collect.py --service=`hostname` --format=falcon --falcon-addr=http://127.0.0.1:1988/v1/push >/dev/null 2>&1
        sleep 2s
done

3. 采集一段时间后，在 falcon 的 dashboard 浏览监控数据:
   

我的几点疑问：

1. 由上述 1 脚本可以看出，每分钟访问 /monitor/basic_status、/hello 和 /monitor/nginx_status 接口至少有几十次，但在 dashboard 都显示每分钟 只1次？这明显跟 README.md 上截图的监控效果不一样。
2. 我的使用方式是否存在误区？如果有，请指正。

@GuyCheung 你好，我在open-falcon中配置了latency_xx的策略，但是当达到了配置的阀值时，alarm没有报警。