You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Piotr Praczyk <pi...@gmail.com> on 2009/06/09 13:51:47 UTC

Starting/stopping hadoop

Hi

I encountered a very peculiar behavior of my Hadoop cluster

When I try to stop it, it claims not to be started

test5.dev ~ # stop-all.sh
no jobtracker to stop
test6.dev: no tasktracker to stop
test5.dev: no tasktracker to stop
no namenode to stop
test5.dev: no datanode to stop
test6.dev: no datanode to stop
test5.dev: no secondarynamenode to stop


(the same comes for HBase)

In the same time, I am able to submit the tasks, access hdfs, hbase.
Do you have any idea, what can be the reason of such state ?
I am running the scripts from the same node, I started hadoop from


cheers
Piotr

hadoop-streaming for network simulation

Posted by Simon Lorenz <si...@lorenz-immobilien.net>.
Hi,

maybe somebody could help me with this. 

What I want to do is using Hadoop Streaming for just executing the same
program with different parameters.

I'm using the network simulatioin software Omnet++ and I want to run
this simulation in parallel.

Omnet Programs can be executed from a linux shell, the just need an
omnetpp.ini file for configuration.
So first step for me is, to get omnet running on hadoop with a simple
example and the same parameters.

So which hadoop-paramters I have to use when I start Hadoop Streaming?
(I think hadoop streaming is the right way to do this, or not?)

Acutally I try something like this, but the streaming job fails.

bin/hadoop jar hadoop-0.18.3-streaming.jar 
-input /input/fifo  
-output /output/fifo 
-mapper "/home/simon/omnetpp-4.0/samples/fifo/fifo -u Cmdenv -c Fifo1"
-file /home/simon/omnetpp-4.0/samples/fifo/fifo 
-reducer NONE

So the omnet program is just the mapper.

- in /input/fifo is the omnetpp.ini file located, that omnet fifo
example needs to run the job.

fifo -u Cmdenv -c Fifo1 are the parameters to start omnet without the
graphic interface. But also when I put these parameters into a shell
srcipt and run the script with hadoop I get the same error.

The omnet program is installed on all machines.

Maybe I have to give the omnet.ini file in some other way to the
omnet-fifo program? I don't know.

I'm no programmer and I even don't know if this approuch goes into the
right direction.
For any hints or suggestions I would be very glad.

P.S. Sorry for my bad english

Simon Lorenz
Karlsruhe
Germany


Re: Starting/stopping hadoop

Posted by Bill Graham <bi...@gmail.com>.
This happens when the processes are running, but the PID files either don't
exist or have invalid PIDs in them. Check on these files on each of your
hosts. The location of the PID file is specified in conf/hadoop-env.sh.

On Tue, Jun 9, 2009 at 9:28 AM, HRoger <ha...@163.com> wrote:

>
> May be PID's problem!
>
> Piotr Praczyk wrote:
> >
> > Hi
> >
> > I encountered a very peculiar behavior of my Hadoop cluster
> >
> > When I try to stop it, it claims not to be started
> >
> > test5.dev ~ # stop-all.sh
> > no jobtracker to stop
> > test6.dev: no tasktracker to stop
> > test5.dev: no tasktracker to stop
> > no namenode to stop
> > test5.dev: no datanode to stop
> > test6.dev: no datanode to stop
> > test5.dev: no secondarynamenode to stop
> >
> >
> > (the same comes for HBase)
> >
> > In the same time, I am able to submit the tasks, access hdfs, hbase.
> > Do you have any idea, what can be the reason of such state ?
> > I am running the scripts from the same node, I started hadoop from
> >
> >
> > cheers
> > Piotr
> >
> >
>
> --
> View this message in context:
> http://www.nabble.com/Starting-stopping-hadoop-tp23941355p23946500.html
> Sent from the Hadoop core-user mailing list archive at Nabble.com.
>
>

Re: Starting/stopping hadoop

Posted by HRoger <ha...@163.com>.
May be PID's problem!

Piotr Praczyk wrote:
> 
> Hi
> 
> I encountered a very peculiar behavior of my Hadoop cluster
> 
> When I try to stop it, it claims not to be started
> 
> test5.dev ~ # stop-all.sh
> no jobtracker to stop
> test6.dev: no tasktracker to stop
> test5.dev: no tasktracker to stop
> no namenode to stop
> test5.dev: no datanode to stop
> test6.dev: no datanode to stop
> test5.dev: no secondarynamenode to stop
> 
> 
> (the same comes for HBase)
> 
> In the same time, I am able to submit the tasks, access hdfs, hbase.
> Do you have any idea, what can be the reason of such state ?
> I am running the scripts from the same node, I started hadoop from
> 
> 
> cheers
> Piotr
> 
> 

-- 
View this message in context: http://www.nabble.com/Starting-stopping-hadoop-tp23941355p23946500.html
Sent from the Hadoop core-user mailing list archive at Nabble.com.