matlabpool local files

12 次查看（过去 30 天）
Daozheng Chen
Daozheng Chen 2012-2-26
Hi:
I am using matlabpool for the jobs which I submit to a cluster. This cluster using Sun Grid Engine (SGE) as the scheduler. I am using "matlabpool local 2;" for each submitted job, and this will a set of files in ".matlab/local_scheduler_data/R2011b", such as "Job10.common.mat."
As more jobs get submitted, The job has a higher chance of getting failed. Sometimes, it gots this error:
Error using matlabpool (line 136)
Failed to open matlabpool. (For information in addition to the causing error,
validate the configuration 'local' in the Configurations Manager.)
Caused by:
Error using distcomp.interactiveclient/start (line 88)
Failed to start matlabpool.
This is caused by:
Unable to read MAT-file
/home-nfs/dzchen/.matlab/local_scheduler_data/R2011b/Job19/Task1.in.mat
File may be corrupt.
Any suggestion is appreciated!
Thank you!
Daozheng
  1 个评论
Dana
Dana 2012-9-25
We are also having intermittent problems when launching multiple instances of matlab on a compute nodes of our cluster. We get this error sometimes and a variety of other errors that seem to be related to files in the .matlab folder. Was this issue ever resolved?

请先登录，再进行评论。

请先登录，再回答此问题。

采纳的回答

Thomas
Thomas 2012-9-25
Did you validate your SGE configuration? Do you have the SGE-specific submit files on the path? These submit function files are required to submit SGE commands to the queueing system.
We have an SGE scheduler with MATLAB DCS which works pretty well.
THe files for R2012a are:
SGEClusterInfo.m getJobStateFcn.m
communicatingJobWrapper.sh getRemoteConnection.m
communicatingSubmitFcn.m getSubmitString.m
createSubmitScript.m independentJobWrapper.sh
deleteJobFcn.m independentSubmitFcn.m
extractJobId.m
The files for R2011 b and older are
createSubmitScript.m
destroyJobFcn.m
distibsubmit.m
distributedJobWrapper.sh
distributedSubmitFcn.m
extractJobId.m
SGEClusterInfo.m
submitjob.m
getJobStateFcn.m
getRemoteConnection.m
getSubmitString.m
parallelJobWrapper.sh
parallelSubmitFcn.m
These files should be availabel from the Mathworks.. Please follow the readme that come along with these files as you will have to edit some parts of these files to suit your SGE cluster queueing system..
  1 个评论
Jason Ross
Jason Ross 2012-9-25
These files are available in matlabroot\toolbox\distcomp\examples, and go over configuring the generic scheduler interface.

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Job and Task Creation 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by