Also you could also just use the prebuilt binaries for those versions that are available from the asf download sites to keep it simple.
Ashish ________________________________ From: Zheng Shao [mailto:[email protected]] Sent: Thursday, May 21, 2009 2:59 AM To: [email protected] Subject: Re: jobtracker deadlock immediately when I launch hive query This problem is mostly likely to be caused by inconsistent hadoop versions. Which version of hadoop do you use on the cluster? Make sure you run "ant clean package -Dhadoop.version=0.18.3' where 0.18.3 is your hadoop version on the cluster. Let us know what hadoop versions you are using, and whether that helps. Zheng On Thu, May 21, 2009 at 2:54 AM, 罗李 <[email protected]<mailto:[email protected]>> wrote: Hi all: I got a problem when I use Hive release-0.3.0-rc1 on top of Hadoop-0.19.1, When I followed the HiveGetingStarted to the step of " SELECT a.foo FROM invites a WHERE a.ds='<DATE>'; " I got no response from hadoop, and the jobtracker seem abnormal when I try to fresh the page of jobtracker.jsp. I use the jstack tool to get the stack information of the jobtracker process,and I got the follow: Found one Java-level deadlock: ============================= "SocketListener0-7": waiting to lock monitor 0x0000002c16f420d0 (object 0x0000002a9eedde60, a org.apache.hadoop.mapred.JobTracker), which is held by "IPC Server handler 6 on 9001" "IPC Server handler 6 on 9001": waiting to lock monitor 0x0000002c1a5dfdc0 (object 0x0000002aa050ce88, a org.apache.hadoop.mapred.JobInProgress), which is held by "initJobs" "initJobs": waiting to lock monitor 0x0000002c16f420d0 (object 0x0000002a9eedde60, a org.apache.hadoop.mapred.JobTracker), which is held by "IPC Server handler 6 on 9001" Everytime I try to launch a hive query ,this situation happened. If I switch to the trunk version of hive, this situation disappeared. Please help me ! luoli <http://svn.apache.org/repos/asf/hadoop/hive/tags/release-0.3.0-rc0/> -- Yours, Zheng
