I met the same problem on window+cygwin, don't know the root cause, but it
can be worked around by a explicit "mapred.child.tmp" property. Try add
these lines to your /mapred-site.xml:
mapred.child.tmp
/tmp/hadoop/mapred/mapred.child.tmp
The examples then worked on my laptop.
Be
1) only the namenode is "formatted" and what happens is basically the
image file is created and prepped. The image file holds the meta data
about how your files are stored on the cluster.
2) The datanodes are not formatted in the conventional sense. Their
(datanode) disk usage will grow only wh
These questions are usually answered once you start using the system but
I'll provide some quick answers.
1. Hadoop uses the local file system at each node to store blocks. The only
part of the system that needs to be formatted is the namenode which is where
Hadoop keeps track of the logical H
I've already tried that. Are you using 0.20.2?
Thank you,
Carlos Eduardo
On Sun, May 9, 2010 at 8:34 AM, Yang Li wrote:
> I met the same problem on window+cygwin, don't know the root cause, but it
> can be worked around by a explicit "mapred.child.tmp" property. Try add
> these lines to your /
Dear Jimmy and Chris:
I am reading your book (thank you for providing the pre-release version) and
I find it great in contents and in style. Thank you!
Sincerely,
Mark
On Sat, May 8, 2010 at 1:25 PM, Jimmy Lin wrote:
> Hi everyone,
>
> I'm pleased to announce the publication a new book on MapR
Hi,
I have class that I run on the master node and submits
a bunch of MR jobs to my cluster but how can I tell where
each job actually executed? I'm using Cloudera's 0.20.2+228.
I don't see any commands or pages in the GUI that tell me this.
Looks like there are some classes that might provide
InetAddress.getLocalHost() should give you the hostname for each
mapper/reducer
On Sun, May 9, 2010 at 12:16 PM, Alan Miller wrote:
> Hi,
>
> I have class that I run on the master node and submits
> a bunch of MR jobs to my cluster but how can I tell where
> each job actually executed? I'm using
Hey Michael,
The script specified by dfs.network.script is passed both host names
and IPs. In most cases an IP is passed, however in some cases (eg when
using dfs.hosts files) a hostname is passed.
Thanks,
Eli
ps - useful pointers:
http://mail-archives.apache.org/mod_mbox/hadoop-common-user/200
Yes, I'm using 0.20.2. I run it in pseudo mode on windows xp + cygwin as
a private playground. All default settings with following change:
core-site.xml:
-
fs.default.name
hdfs://localhost:9000
hadoop.tmp.dir
/tmp/hadoop
hdfs-site.xm
Do you mean where each task run ? You can look at the job tracker web
ui where you can find each job's status and info.
On Mon, May 10, 2010 at 3:16 AM, Alan Miller wrote:
> Hi,
>
> I have class that I run on the master node and submits
> a bunch of MR jobs to my cluster but how can I tell wher
10 matches
Mail list logo