[maker-devel] Forks.pm error when running maker with dsindex

[Carson Holt]

It's better to run fewer jobs with more cpus given to MPI rather than many
jobs with few cpus (i.e. mpiexec -n 4).

To correct errors, you just restart MAKER.  No need to set the -a flag
unless you want to rerun everything, and not just the failed contigs.

--Carson


On 8/6/14, 3:03 AM, "Jeanne Wilbrandt" <j.wilbrandt at zfmk.de> wrote:

>
>Hi!
>
>Yes, we are running 20 jobs simultaneously, almost, i.e., as much as our
>cluster can
>take. Do you think this is too much?
>
>Please find attached the output file (containing the STDERR) of the
>dsindex-run, and one
>example output of one of the pieces.
>
>Another quick question to make sure I understood the guides correctly: If
>a job did not
>finish properly, it should suffice to restart the same thing just with
>the -a flag and it
>should clean up and finish what it was supposed to, right? (i.e., it's
>not necessary to
>trace and delete the unfinished output manually?)
>
>Thank you again!
>Jeanne Wilbrandt
>
>zmb // ZFMK // University of Bonn
>
>
>
>On 08/05/2014 08:00 PM, maker-devel-request at yandell-lab.org wrote:
>>
>>
>>    1. Re: Forks.pm error when running maker with dsindex (Carson Holt)
>>
>>
>> ----------------------------------------------------------------------
>>
>> Message: 1
>> Date: Mon, 04 Aug 2014 14:27:08 -0600
>> From: Carson Holt <carsonhh at gmail.com>
>> To: Jan Philip Oeyen <jp.oeyen at uni-bonn.de>,
>> 	<maker-devel at yandell-lab.org>
>> Subject: Re: [maker-devel] Forks.pm error when running maker with
>> 	dsindex
>> Message-ID: <D005484F.DFD2%carsonhh at gmail.com>
>> Content-Type: text/plain; charset="utf-8"
>>
>> Sorry for the slow reply.  I was on vacation all last week.  Do you
>>have the
>> full STDERR? sometimes the last error is irrelevant and it's just the
>>result
>> of a failure further upstream. Also are you running 20 independent maker
>> jobs simultaneously?
>>
>> --Carson
>>
>>
>> From:  Jan Philip Oeyen <jp.oeyen at uni-bonn.de>
>> Date:  Monday, July 28, 2014 at 6:22 AM
>> To:  <maker-devel at yandell-lab.org>
>> Subject:  [maker-devel] Forks.pm error when running maker with dsindex
>>
>> Hi all,
>> we are currently having some unexpected errors when running maker on a
>> genome which is split in several parts. Our cluster admin reported the
>> following error message:
>>
>> Argument "ALRM" isn't numeric in exit at /share/scientific_bin/perlmodu
>> les/lib/site_perl/5.14.2/x86_64-linux-thread-multi/forks.pm
>> <http://forks.pm>  line 2188.
>> SIGTERM received
>> SIGTERM received
>> SIGTERM received
>>
>> We were using maker with the '-g' option on a single genome which is
>>split
>> into 20 parts, where 19 parts are equally large and the last contains
>>about
>> 20 sequences more. After that we ran Maker using dsindex to clean up the
>> output. We are currently using maker v2.31 on 4 threads and forks v0.34.
>>
>> If any further info is needed to clarify the problem, please let me
>>know and
>> I will provide as much as possible.
>>
>> Thank you for your help!
>>
>> Best regards,
>> Jan Philip Oeyen
>> ZFMK // ZMB // University of Bonn
>>