I'm glad to report back that both of my fetches, one using Hadoop 7.1 and the other 9.1 were successful. It looks like the old Hadoop jar file was picked up by the java process before it was completely removed by the system. There are a few things that could have caused that, but none of them are related to Nutch itself. I thank you for your assistance, and my next little project is to get Hadoop's native lib package to compile under BSD so that I may bring its compression ability to Nutch. Sean
----- Original Message ---- From: Andrzej Bialecki <[EMAIL PROTECTED]> To: [email protected] Sent: Thursday, December 14, 2006 7:00:46 AM Subject: Re: error with trunk: linkdb copied to wrong dir Sean Dean wrote: > I'm pretty sure there wasn't, as I don't have any older hadoop*.jar files in > the classpath. The only thing that comes to mind is if an older hadoop file, > which would most likely be hadoop-0.7.1.jar was loaded into memory before it > was deleted/replaced with hadoop-0.9.1.jar. To make certain, I have also > started another fetch, using the latest trunk and same segment on another > system with no previous nutch builds, so its all fresh. This should take <24 > hours to complete, and when its done I will report back with its status and > any further details if available. > If you can, it would be better to create a simple db with one url, generate a segment and fetch it - if it's a wrong version of hadoop then symptoms should be the same, no matter what the number of urls. -- Best regards, Andrzej Bialecki <>< ___. ___ ___ ___ _ _ __________________________________ [__ || __|__/|__||\/| Information Retrieval, Semantic Web ___|||__|| \| || | Embedded Unix, System Integration http://www.sigram.com Contact: info at sigram dot com
