Abhijit, Double check that you can inject the url's folder on non dfs, make sure url's are utf-8 compliant. Hit that error with a non utf-8 compliant "windows edited" url file. If that's not it, make sure you've set the dfs as the default working directory in the configs. -Jay Pound
-----Original Message----- From: Dennis Kubes [mailto:[EMAIL PROTECTED] Sent: Sunday, May 25, 2008 7:46 AM To: [email protected] Subject: Re: Please help me get Nutch working Ok, then please provide more information, logs, etc. if you would like more help with your problem. Dennis Abhijit Bera wrote: > Dennis: > > I have uploaded my urls directory to the DFS :) If I didn't upload my > URLS directory, Nutch wouldn't work, it would throw up an error. > > On Fri, 2008-05-23 at 07:41 -0500, Dennis Kubes wrote: >> It could be any number of things. Without more information, best guess >> is that you are not uploading the url directory to the dfs before >> running inject. >> >> Dennis >> >> Abhijit Bera wrote: >>> Hi >>> >>> I have set up Nutch on a 4 Node cluster of Ubuntu Boxes each having >>> different h/w configs. >>> >>> I have setup Hadoop on this cluster and it works perfectly with the word >>> count example supplied. >>> >>> But some how I'm not able to get Nutch to execute correctly. I want to >>> learn how it works on a Hadoop cluster so that I can start developing on >>> Nutch but I'm always getting this error: >>> >>> Error: Generator: 0 records selected for fetching, exiting ... >>> >>> I used this command to start the crawl: >>> bin/nutch crawl urls -dir crawled -depth 10 >>> >>> I have 20 urls which I have specified for crawling I tried formatting >>> the namenode several times and recrawling but I still get the same >>> error. >>> >>> I'll send my config files as an tar attachment. Please tell me where I >>> have made a mistake. >>> >>> Thanks >>>
