Michael Thomas wrote:
Hey guys,
During the SC09 exercise, our data transfer tool was using the FUSE
interface to HDFS. As Brian said, we were also reading 16 files in
parallel. This seemed to be the optimal number, beyond which the
aggregate read rate did not improve.
We have worked sched
For security reasons I am required to conform and use a different S3 library
that I am provided to access S3 data. If I write an adapter against the
native file system store class to access S3 using my own library, do I still
get the same benefits that I would get for using the default file system
Correct. The master runs the namenode and jobtracker, but not a
datanode or tasktracker.
Tom
On Tue, Nov 24, 2009 at 4:57 PM, Mark Kerzner wrote:
> Hi,
>
> do I understand it correctly that, when I launch a Hadoop cluster on EC2,
> the master will not be doing any work, and it is just for organi
On Tue, Nov 24, 2009 at 9:27 PM, Mark Kerzner wrote:
> Yes, Tom, I saw all these problems. I think that I should stop trying to
> imitate EMR - that's where the storing data on S3 appeared, and transfer
> data directly to the Hadoop cluster. Then I will be using all as intended.
>
> Is there a way
Just FYI, Hadoop and M/R is a distributed computing system. So, there
is a problem of locality and location of sub-matrix blocks. Moreover,
M/R iteration method is really slow.
To perform the matrix multiplication (and also graph algorithm) on
Hadoop, Apache Hama team is considering a BSP (bulk sy
Hi Gimich,
Could you describe your matrix multiplication problem in more details? Are the
matrices sparse or dense? How big is the on-disk-size of a matrix?
Thanks.
Nicholas Sze
- Original Message
> From: Edward J. Yoon
> To: common-user@hadoop.apache.org
> Sent: Tue, November 24
I have 2 clusters:
30 nodes running 0.18.3
and
36 nodes running 0.20.1
I've intermittently seen the following errors on both of my clusters, it
happens when writing files.
I was hoping this would go away with the new version but I see the same
behavior on both versions.
The namenode logs don't
Hi,
I have started the Apache distribution of hadoop-0.19, and I noticed that
this Fedora image has only 3 Gig hard drive. Why is it so small? It was 50%
on startup, and with me adding a few things it is now at 80%. Anything I can
do to increase this? I would not have much room to work on.
Thank
You could check whats going on on second datanode (10.1.75.104). The connect
to it from the first datanode took longer than 2 min.
Btw, it looks to be a bug that there is a mismatch in timeout used by
datanode and the client. Ideally client's timeout should be larger than the
timeout used by the d
Hi Mark,
The root partition is small, but there is plenty of storage on the
/mnt partition. See http://aws.amazon.com/ec2/instance-types/.
Cheers,
Tom
On Wed, Nov 25, 2009 at 12:30 PM, Mark Kerzner wrote:
> Hi,
>
> I have started the Apache distribution of hadoop-0.19, and I noticed that
> this
Tom,
Can I install applications on /mnt? Are they saved on reboot and on image
bundling? If not, I will have to be extra tight about this, but use /mnt as
my working area. That is still something, so thank you.
Mark
On Wed, Nov 25, 2009 at 2:48 PM, Tom White wrote:
> Hi Mark,
>
> The root part
Oh, and all instances http://aws.amazon.com/ec2/instance-types/ have 10 Gig
root partition, so where did this go? I guess somebody wanted extra small
root.
On Wed, Nov 25, 2009 at 2:48 PM, Tom White wrote:
> Hi Mark,
>
> The root partition is small, but there is plenty of storage on the
> /mnt
Hi,
I get this part-0.deflate instead of part-0.
How do I get rid of the deflate option?
Thank you,
Mark
Hello,
I have a set of input files part-r-* which I will pass through another
map(no reduce). the part-r-* files consist of key, values, keys being
small, values fairly large(MB's)
I would like to index these, i.e run a map, whose output is key and
/filename/ i.e to which part-r-* file the partic
Hi,
".deflate" is the default compression codec used when parameter to generate
compressed output is true ( mapred.output.compress ).
You may set the codec to be used via mapred.output.compression.codec, some
commonly used are available in hadoop.io.compress package...
Amogh
On 11/26/09 11:03
For testing purposes you can also try to disable the compression:
conf.setBoolean("mapred.output.compress", false);
Then you can look at the output.
- tim
Amogh Vasekar wrote:
Hi,
".deflate" is the default compression codec used when parameter to generate
compressed output is true ( mapred.o
Conf.get(map.input.file) is what you need.
Amogh
On 11/26/09 12:35 PM, "Saptarshi Guha" wrote:
Hello,
I have a set of input files part-r-* which I will pass through another
map(no reduce). the part-r-* files consist of key, values, keys being
small, values fairly large(MB's)
I would like to
Thank you.
Regards
Saptarshi
On Thu, Nov 26, 2009 at 2:10 AM, Amogh Vasekar wrote:
> Conf.get(map.input.file) is what you need.
>
> Amogh
>
>
> On 11/26/09 12:35 PM, "Saptarshi Guha" wrote:
>
> Hello,
> I have a set of input files part-r-* which I will pass through another
> map(no reduce). the
Hello again,
I'm using Hadoop 0.21 and its context object e.g
public void setup(Context context) {
Configuration cfg = context.getConfiguration();
System.out.println("mapred.input.file="+cfg.get("mapred.input.file"));
displays null, so maybe this fell out by mistake in the api change?
R
19 matches
Mail list logo