[maker-devel] Maker MPICH2 issue (Multiple MAKER processes have been started in the same directory)

[Barry Moore]

Hi Marc,

Thanks so much for sharing the results of all your hard work!  We will follow up on your feedback and incorporate the necessary updates to MAKER docs and installation procedures.

B

On Oct 26, 2013, at 5:54 AM, Marc P. Hoeppner wrote:

> Dear List,
> 
> after much more debugging, I found the solution myself. Here are the details:
> 
> The Maker installation instructions refer to 'mpich2' in multiple places, including weblinks (whchi are outdated btw). Some digging revealed that mpich2 has since been renamed and merged back into mpich. So the website is www.mpich.org and the current stable version is mpich-3.0.4.  Closely following the installation instruction yielded a working copy of mpiexec that Maker was willing to accept. Good news - but I suggest to update the documentation. Most importantly for either SL/CentOS users, mpich2 as packaged with e.g. Scientific Linux *does not* work with Maker. You will need to download the latest stable build and compile it yourself (the mpich2 version bundled with Maker doesn't work either btw).
> 
> Cheers,
> 
> Marc
> 
>> Dear list,
>> 
>> I know that MPICH2 related issues have been discussed previously, but the suggested solutions don't seem to work for me.
>> Briefly, I have set up a new cluster, with a fresh install of mpich2 (SL 6.4, from the yum repos) and a newly compiled build of Maker 2.28.
>> 
>> When I start maker with mpiexec, I get spammed with
>> 'WARNING: Multiple MAKER processes have been started in the same directory.'
>> 
>> Here is what I have tried so far - still getting the error :-< Any suggestions would be appreciated
>> 
>> --------
>> MPICH2
>> ---------
>> 
>> which mpiexec:
>> /usr/lib64/mpich2/bin/mpiexec
>> 
>> Started the mpd ring as normal user across 5 nodes (bnode-01 to bnode-04, and the head node bhead)
>> 
>> mpdtrace:
>> bhead
>> bnode-02
>> bnode-03
>> bnode-04
>> bnode-01
>> 
>> mpdringtest:
>> time for 1 loops = 0.00128293037415 seconds
>> 
>> mpixec -n 32 -machinefile hostfile hostname:
>> 
>> <long list of host names, as ecpected>
>> 
>> I also ran the Skampi Mpi Benchmark, which ran through just fine.
>> 
>> -----------
>> MAKER
>> -----------
>> 
>> ./Build status
>> 
>> PERL Dependencies:      VERIFIED
>> External Programs:      VERIFIED
>> External C Libraries:   VERIFIED
>> MPI SUPPORT:            ENABLED
>> MWAS Web Interface:     DISABLED
>> MAKER PACKAGE:          CONFIGURATION OK
>> 
>> During configuration/installation, Build.PL automatically detected all MPI data and never complained, so I have to assume that everything went fine there.
>> 
> 
> 
> -- 
> ----------
> Marc P. Hoeppner, PhD
> Department of Microbiology and Medical Biochemistry
> Uppsala University, Sweden
> marc.hoeppner at imbim.uu.se
> 
> 
> _______________________________________________
> maker-devel mailing list
> maker-devel at box290.bluehost.com
> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org

Barry Moore
Research Scientist
Dept. of Human Genetics
University of Utah
Salt Lake City, UT 84112
--------------------------------------------
(801) 585-3543




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20131026/ffe6e917/attachment-0003.html>