I'm guessing this is http://issues.apache.org/jira/browse/HADOOP-3139 which is fixed in Hadoop 0.16.3 (currently being voted on for release by the developer community).

Nige

On Apr 14, 2008, at 3:07 AM, Andreas Kostyrka wrote:
As another item, the submitting Java process hangs in a futex call:

[EMAIL PROTECTED]:~# strace -p 3810
Process 3810 attached - interrupt to quit
futex(0xb7d6ebd8, FUTEX_WAIT, 3832, NULL

and hangs, hangs, hangs, ...

Andreas


Am Montag, den 14.04.2008, 11:46 +0200 schrieb Andreas Kostyrka:
Ok, a short grep in the sources suggests that the exceptions happen just
in the closeAll method of FileSystem. So no indication what hadoop is
working on :(


Am Montag, den 14.04.2008, 07:26 +0200 schrieb Andreas Kostyrka:
Hi!

I'm getting the following hang, when trying to run a streaming command:

[EMAIL PROTECTED]:~/hadoop-0.16.2$ time bin/hadoop jar contrib/streaming/hadoop-0.16.2-streaming.jar -mapper '/home/ hadoop/bin/llfp -f [EMAIL PROTECTED] -t [EMAIL PROTECTED] -s heaven.kostyrka.org -d gen_dailysites -d fb_memberfind ' -reducer NONE -input run-0003-input/* -output run-0003-output
additionalConfSpec_:null
null=@@@userJobConfProps_.get(stream.shipped.hadoopstreaming
packageJobJar: [/tmp/hadoop-hadoop/hadoop-unjar60050/] [] /tmp/ streamjob60051.jar tmpDir=null 08/04/14 05:22:10 INFO fs.FileSystem: FileSystem.closeAll() threw an exception: org.apache.hadoop.io.MultipleIOException: 2 exceptions [java.io.IOException: S3FileSystem ([EMAIL PROTECTED]) and Key ([EMAIL PROTECTED]://lokgad) do not match., java.io.IOException: LocalFileSystem([EMAIL PROTECTED]) and Key([EMAIL PROTECTED]://null) do not match.]


The exception happens only when I press Control-C after many minutes.

The cluster uses an HDFS/S3 filesystem, and concurrently to this I'm
issueing many many "relative small" hadoop fs -put requests, copying in
data to a different folder.

The above hang happened multiple times, so I wonder what could it be?

Andreas

Reply via email to