[maker-devel] Problems with openMPI in multiple computing nodes

[Carson Holt]

You don’t have berkleyDB installed on your system, so BioPerl is trying to fall back to another index format that has issues on network mounted file systems. You can try and install BerkleyDB then the related perl module (https://metacpan.org/pod/BerkeleyDB <https://metacpan.org/pod/BerkeleyDB>). You would then need to reinstall BioPerl and MAKER. You can also try running on a single CPU until indexing finishes, then launch MAKER.  That might be enough to get around any early race conditions.

—Carson


> On Apr 26, 2020, at 12:58 AM, Xu, taosheng <taosheng.x at gmail.com> wrote:
> 
> Hello, 
> I am using a  computer cluster with 20 nodes(40cpus per node) for gene annotation. I submit my maker task to one node with 40 CPUs using openMPI. Everything is well. 
> But I encounter the problem when submitting the same maker task to the cluster with multiple nodes (120 cpus) There are errors shown below.  
> I would also appreciate any advice. Thank you.
> 
> Best regards,
> Taosheng
> 
> 
> STATUS: Processing and indexing input FASTA files...
> cannot remove directory for home/20200425/genome.maker.output/mpi_blastdb/te_proteins%2Efasta.mpi.10//.dbtmp0: No such file or directory at /maker/bin/../lib/FastaDB.pm line 145.
> cannot remove directory for /home/20200425/genome.maker.output/mpi_blastdb/te_proteins%2Efasta.mpi.10//.dbtmp0: Directory not empty at /maker/bin/../lib/FastaDB.pm line 145.
> cannot remove directory for /home/20200425/genome.maker.output/mpi_blastdb/te_proteins%2Efasta.mpi.10//.dbtmp0: Directory not empty at /maker/bin/../lib/FastaDB.pm line 145.
> _______________________________________________
> maker-devel mailing list
> maker-devel at yandell-lab.org
> http://yandell-lab.org/mailman/listinfo/maker-devel_yandell-lab.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20200526/5c4c7cca/attachment-0003.html>