由於Google Cloud Logging是透過fluentd實作,加上fluentd介接BigQuery也非常簡單...
有這個機會跟fluentd多親近一下... :D
在玩了fluentd一陣子後,雖然他提供了不少的input, output, filter plugin,但是總是有一些無法滿足的地方
加上fluentd是ruby friendly的系統...@@,沒學過ruby總是難以在上面發揮...@@
看了一陣子,終於下手測試output exec plugin... 期待可以藉由他來用第三方程式語言去介接其他output plugin所未提供的服務...
下面是一段將syslog source輸入,然後output到node.js程式的範例config
<source>type syslogprotocol_type udpport 514bind 0.0.0.0tag system</source><source>type syslogprotocol_type tcpport 514bind 0.0.0.0tag system</source><match **>type copy<store>type stdoutoutput_type json</store><store>type execcommand /opt/node/bin/node /home/simonsu/data/test.jsformat jsontime_format %Y-%m-%d %H:%M:%Sflush_interval 5sbuffer_path /tmp</store></match>
然後其中"/home/simonsu/data/test.js"則是我們預計要執行的node.js程式
在exec中,以node.js會在指令列接收到類似:/tmp.20150611.q5183cc42694434d7.log的字串
該字串是buffer中未flush的資料檔,以上面設定而言,資料會以json的方式收錄
內容長得像這樣:
{"host":"simon-centos6","ident":"run-parts","message":"(/etc/cron.hourly)[2050 starting 0anacron"}{"host":"simon-centos6","ident":"anacron","pid":"2061","message":"Anacron started on 2015-06-11"}{"host":"simon-centos6","ident":"run-parts","message":"(/etc/cron.hourly)[2063 finished 0anacron"}{"host":"simon-centos6","ident":"anacron","pid":"2061","message":"Will run job `cron.daily' in 17 min."}{"host":"simon-centos6","ident":"anacron","pid":"2061","message":"Will run job `cron.weekly' in 37 min."}{"host":"simon-centos6","ident":"anacron","pid":"2061","message":"Will run job `cron.monthly' in 57 min."}{"host":"simon-centos6","ident":"anacron","pid":"2061","message":"Jobs will be executed sequentially"}
先不討論關於接收到檔案後要做的事情,我們可以直接在test.js中透過fs把該檔案內容讀出來:
var fs = require('fs');var txt = fs.readFileSync(process.argv[2],'utf8');
//ex: console.log('[OUT]', txt);
後續可以針對txt來做後續的處理動作...
附註:如果處理的當下有exception,fluentd會將內容暫存在buffer_path內(也就是前面所接到的參數)
接下來啟動測試,透過fluentd或td-agent執行(ex: td-agent -c xxx.conf)後,如果node.js程式當中有做console.log,則在console output內可以看到console.log的結果喔!
留言
張貼留言