Prev: Using COM in Matlab
Next: pixel list from bounding box?
From: Jerry on 22 Jul 2008 14:02

I am using Matlab R2008a on a Lenovo X60 Tablet with Dual
Core. I have a problem to start "matlabpool." For example,
if I run the following test code:

clc; clear all;
matlabpool local 2
parfor i = 1:8
A(i) = i;
end
A

It will give me an error message

Starting matlabpool using the parallel configuration 'local'.
Waiting for parallel job to start...
Performing parallel job cleanup...
Done.
??? Error using ==>
distcomp.interactiveclient.pGetSockets>iThrowIfBadParallelJobStatus
at 107
The interactive parallel job finished without any messages.

Error in ==> matlabpool at 90
client.start('matlabpool', numlabs, config, 'nogui');

Error in ==> pcode at 6
matlabpool local 2

What is the problem with this? Thanks.

Jinhui
From: Edric M Ellis on 23 Jul 2008 03:29
"Jerry " <jinhui_bai(a)hotmail.com> writes:

> I am using Matlab R2008a on a Lenovo X60 Tablet with Dual
> Core. I have a problem to start "matlabpool." For example,
> if I run the following test code:
>
> clc; clear all;
> matlabpool local 2
> parfor i = 1:8
> A(i) = i;
> end
> A
>
> It will give me an error message
>
> Starting matlabpool using the parallel configuration 'local'.
> Waiting for parallel job to start...
> Performing parallel job cleanup...
> Done.
> ??? Error using ==>
> distcomp.interactiveclient.pGetSockets>iThrowIfBadParallelJobStatus
> at 107
> The interactive parallel job finished without any messages.

Hi,

There are a few things that can cause a matlabpool to fail to start up - the
most common being pathdef.m or startup.m changes that mean that the MATLAB
workers don't have the correct path set up. One other symptom of this sort of
thing is that if you check in task manager, you may see some extra MATLAB
processes running that aren't doing anything. If you've modified your pathdef.m,
you should try reverting that.

Another debug stage you could try is the following:

---
s = findResource( 'scheduler', 'Type', 'local' );
j = s.createParallelJob( 'Max', 2 );
j.createTask( @labindex, 1 );
j.submit
j.wait
s.getDebugLog( j )
---

The "debug log" output may give some indication as to what went wrong. If the
"j.wait" line takes a really long time (2 minutes or more), you should be able
to CTRL-C out of that, and execute the "getDebugLog" line anyway.

Cheers,

Edric.
From: Jinhui Bai on 24 Jul 2008 16:20

Thanks for the suggestion. I found out the problem. It is a
known bug of Parallel Computing Toolbox described in
http://www.mathworks.com/support/bugreports/details.html?rp=454732

I called Mathworks and they fixed my problem by changing my
license file.

Jinhui

Edric M Ellis <eellis(a)mathworks.com> wrote in message
<ytw1w1l83e6.fsf(a)uk-eellis-deb4-64.mathworks.co.uk>...
> "Jerry " <jinhui_bai(a)hotmail.com> writes:
>
> > I am using Matlab R2008a on a Lenovo X60 Tablet with Dual
> > Core. I have a problem to start "matlabpool." For example,
> > if I run the following test code:
> >
> > clc; clear all;
> > matlabpool local 2
> > parfor i = 1:8
> > A(i) = i;
> > end
> > A
> >
> > It will give me an error message
> >
> > Starting matlabpool using the parallel configuration
'local'.
> > Waiting for parallel job to start...
> > Performing parallel job cleanup...
> > Done.
> > ??? Error using ==>
> >
distcomp.interactiveclient.pGetSockets>iThrowIfBadParallelJobStatus
> > at 107
> > The interactive parallel job finished without any messages.
>
> Hi,
>
> There are a few things that can cause a matlabpool to fail
to start up - the
> most common being pathdef.m or startup.m changes that mean
that the MATLAB
> workers don't have the correct path set up. One other
symptom of this sort of
> thing is that if you check in task manager, you may see
some extra MATLAB
> processes running that aren't doing anything. If you've
modified your pathdef.m,
> you should try reverting that.
>
> Another debug stage you could try is the following:
>
> ---
> s = findResource( 'scheduler', 'Type', 'local' );
> j = s.createParallelJob( 'Max', 2 );
> j.createTask( @labindex, 1 );
> j.submit
> j.wait
> s.getDebugLog( j )
> ---
>
> The "debug log" output may give some indication as to what
went wrong. If the
> "j.wait" line takes a really long time (2 minutes or
more), you should be able
> to CTRL-C out of that, and execute the "getDebugLog" line
anyway.
>
> Cheers,
>
> Edric.

 | 
Pages: 1
Prev: Using COM in Matlab
Next: pixel list from bounding box?