nutch 分布式索引（爬虫)

其实，全网抓取比intranet区别再于，

  前者提供了较为多的urls入口，

  没有使用crawl-urlfilter.txt 中并没有限制哪些url ，(如果没有使用crawl命令）

  并通过逐步处理的方式得以可按的局面；

 

在1.3，还有此区别，

  如默认的fetcher.parse是false，使得每次fetch后必须有一个parse step，刚开始老是看不懂为什么tutorial中这样做。。

  其次是，此版本不再有crawl-urlfiter.txt，而是用regex-urlfilter.txt替换。

 

在recrawl时的区别见nutch 数据增量更新

 

其实这个过程可以说是nutch对hadoop的利用的最深体会，我是这样认为的。想想看，当初，hadoop是内嵌在Nutch中，作为其中的一个功能模块。现在版本的nutch将hadop分离出来，但在分布式抓取时又得将它（配置文件，jar等）放回Nutch下。刚开始时老想nutch怎样结合hadoop进行分布式抓取；但分布式搜索还是有些不一样的，因为后者即使也是分布式，但它利用的hdfs对nutch是透明的。

 

install processes:

a.configure hadoop to run on cluster mode;

b.put all the config files belong hadoop(master and slaves) to conf dir of nutch(s) respectively;

c.execute the crawl command (SHOULD use individual commands to do INSTEAD OF 'craw',as 'crawl' is used for intranet usually)

 

here are the jobs belong this step:

Available Jobs
Job tracker Host Name	Job tracker Start time	Job Id	Name	User
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0001	inject crawl-url	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0002	crawldb crawl/dist/crawldb	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0003	generate: select from crawl/dist/crawldb	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0004	generate: partition crawl/dist/segments/2011110720	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0005	fetch crawl/dist/segments/20111107205746	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0006	crawldb crawl/dist/crawldb(update db actually )	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0007	linkdb crawl/dist/linkdb	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0008	index-lucene crawl/dist/indexes	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0009	dedup 1: urls by time	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0010	dedup 2: content by hash	hadoop
master	Mon Nov 07 20:50:54 CST 2011	job_201111072050_0011	dedup 3: delete from index(es)	hadoop

 

 

* the jobs above with same color is ONE step beong the crawl command;

* job 2 :将sortjob結果作为输入(与已有的current数据合并)，生成新的crawldb;所以可以有重复的urls,在reduce中会去重 ?

* job 4:由于存在多台crawlers,所以需要利用partition来划分urls(by host by default),来避免重复让一台机来抓取 ；

 

 

here is the output of resulst:

hadoop@leibnitz-laptop:/xxxxxxxxx$ hadoop fs -lsr crawl/dist/
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00000
-rw-r--r--   2 hadoop supergroup       6240 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00000/data
-rw-r--r--   2 hadoop supergroup        215 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00001
-rw-r--r--   2 hadoop supergroup       7779 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00001/data
-rw-r--r--   2 hadoop supergroup        218 2011-11-07 21:00 /user/hadoop/crawl/dist/crawldb/current/part-00001/index

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:07 /user/hadoop/crawl/dist/index
-rw-r--r--   2 hadoop supergroup        369 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.fdt
-rw-r--r--   2 hadoop supergroup         20 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.fdx
-rw-r--r--   2 hadoop supergroup         71 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.fnm
-rw-r--r--   2 hadoop supergroup       1836 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.frq
-rw-r--r--   2 hadoop supergroup         14 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.nrm
-rw-r--r--   2 hadoop supergroup       4922 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.prx
-rw-r--r--   2 hadoop supergroup        171 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.tii
-rw-r--r--   2 hadoop supergroup      11234 2011-11-07 21:07 /user/hadoop/crawl/dist/index/_2.tis
-rw-r--r--   2 hadoop supergroup         20 2011-11-07 21:07 /user/hadoop/crawl/dist/index/segments.gen
-rw-r--r--   2 hadoop supergroup        284 2011-11-07 21:07 /user/hadoop/crawl/dist/index/segments_2

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000
-rw-r--r--   2 hadoop supergroup        223 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.fdt
-rw-r--r--   2 hadoop supergroup         12 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.fdx
-rw-r--r--   2 hadoop supergroup         71 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.fnm
-rw-r--r--   2 hadoop supergroup        991 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.frq
-rw-r--r--   2 hadoop supergroup          9 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.nrm
-rw-r--r--   2 hadoop supergroup       2813 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.prx
-rw-r--r--   2 hadoop supergroup        100 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.tii
-rw-r--r--   2 hadoop supergroup       5169 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/_0.tis
-rw-r--r--   2 hadoop supergroup          0 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/index.done
-rw-r--r--   2 hadoop supergroup         20 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/segments.gen
-rw-r--r--   2 hadoop supergroup        240 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00000/segments_2
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001
-rw-r--r--   2 hadoop supergroup        150 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.fdt
-rw-r--r--   2 hadoop supergroup         12 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.fdx
-rw-r--r--   2 hadoop supergroup         71 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.fnm
-rw-r--r--   2 hadoop supergroup        845 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.frq
-rw-r--r--   2 hadoop supergroup          9 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.nrm
-rw-r--r--   2 hadoop supergroup       2109 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.prx
-rw-r--r--   2 hadoop supergroup        106 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.tii
-rw-r--r--   2 hadoop supergroup       6226 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/_0.tis
-rw-r--r--   2 hadoop supergroup          0 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/index.done
-rw-r--r--   2 hadoop supergroup         20 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/segments.gen
-rw-r--r--   2 hadoop supergroup        240 2011-11-07 21:04 /user/hadoop/crawl/dist/indexes/part-00001/segments_2

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00000
-rw-r--r--   2 hadoop supergroup       8131 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00000/data
-rw-r--r--   2 hadoop supergroup        215 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00001
-rw-r--r--   2 hadoop supergroup      11240 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00001/data
-rw-r--r--   2 hadoop supergroup        218 2011-11-07 21:01 /user/hadoop/crawl/dist/linkdb/current/part-00001/index

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00000
-rw-r--r--   2 hadoop supergroup      13958 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00000/data
-rw-r--r--   2 hadoop supergroup        213 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00001
-rw-r--r--   2 hadoop supergroup       6908 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00001/data
-rw-r--r--   2 hadoop supergroup        224 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/content/part-00001/index

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00000
-rw-r--r--   2 hadoop supergroup        255 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00000/data
-rw-r--r--   2 hadoop supergroup        213 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00001
-rw-r--r--   2 hadoop supergroup        266 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00001/data
-rw-r--r--   2 hadoop supergroup        224 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_fetch/part-00001/index

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:58 /user/hadoop/crawl/dist/segments/20111107205746/crawl_generate
-rw-r--r--   2 hadoop supergroup        255 2011-11-07 20:58 /user/hadoop/crawl/dist/segments/20111107205746/crawl_generate/part-00000
-rw-r--r--   2 hadoop supergroup         86 2011-11-07 20:58 /user/hadoop/crawl/dist/segments/20111107205746/crawl_generate/part-00001

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_parse
-rw-r--r--   2 hadoop supergroup       6819 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_parse/part-00000
-rw-r--r--   2 hadoop supergroup       8302 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/crawl_parse/part-00001

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00000
-rw-r--r--   2 hadoop supergroup       2995 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00000/data
-rw-r--r--   2 hadoop supergroup        213 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00001
-rw-r--r--   2 hadoop supergroup       1917 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00001/data
-rw-r--r--   2 hadoop supergroup        224 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_data/part-00001/index

drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00000
-rw-r--r--   2 hadoop supergroup       3669 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00000/data
-rw-r--r--   2 hadoop supergroup        213 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00000/index
drwxr-xr-x   - hadoop supergroup          0 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00001
-rw-r--r--   2 hadoop supergroup       2770 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00001/data
-rw-r--r--   2 hadoop supergroup        224 2011-11-07 20:59 /user/hadoop/crawl/dist/segments/20111107205746/parse_text/part-00001/index

 

从以上分析可得，除了merged index外，其它目录都存在两份－对应两台crawlers.

利用这两份索引 ，就可以实现分布式搜索了。

剩下问题：为什么网上介绍的分步方式都没有使用dedup命令？

从  nutch 数据增量更新    上可知， 分布式抓取也应该使用这个dedup命令。

 

see also

http://wiki.apache.org/nutch/NutchTutorial

http://mr-lonely-hp.iteye.com/blog/1075395