[maker-devel] maker 2.22/2.23: Segmentation fault while processing FASTA input
Thomas Hackl
thomas.hackl at uni-wuerzburg.de
Mon Mar 26 04:51:48 MDT 2012
Hi,
I was finally able to "fix" the Problem. I tried everything you
suggested, nice catch on my screwed up Perl lib path btw, but I could
neither install GDBM_File or NDBM_File nor Berkley_DB separately. In the
end I simply removed the ISA relations to the modules in question from
the source code and now it seems to works fine...
sed -i 's/qw(DB_File GDBM_File NDBM_File SDBM_File)/qw(DB_File)/'
So thanks a lot for your prompt and comprehensive response, the debug
executables and the help with the messages.
Regards
Thomas
Am 21.03.2012 23:05, schrieb Carson Holt:
> That is likely the strangest result I could imagine? The --debug option
> really only cause MAKER to print status messages everywhere and nothing
> else, so there must be something else going on between the tests.
>
>
> One things I did notice from the error log though -->
> 0.49 Inline /home/s187512/perl/lib/perl5/site_perl/5.12.1/Inline.pm
> UNKNOWN Inline::denter /home/s187512/perl/lib/perl5/site_perl/5.12.1/Inlin
> e/denter.pm
>
>
> These are both being loaded from perl 5.12.1 (you are using 5.14.1). That
> can cause issues since I know the first one is executed at the C level and
> I wouldn't be surprised if the same is true on the second one.
>
>
>
> I also saw this -->
> Can't locate package GDBM_File for @AnyDBM_File::ISA at
> /storage/tmp/software/perl-5.14.1/lib/5.14.1/x86_64-linux-thread-multi/DB_F
> ile.pm line 293.
> Can't locate package NDBM_File for @AnyDBM_File::ISA at
> /storage/tmp/software/perl-5.14.1/lib/5.14.1/x86_64-linux-thread-multi/DB_F
> ile.pm line 293.
> Can't locate package SDBM_File for @AnyDBM_File::ISA at
> /storage/tmp/software/perl-5.14.1/lib/5.14.1/x86_64-linux-thread-multi/DB_F
> ile.pm line 293.
>
>
>
> This appears could be related to Bio::DB::Fasta (which would make sense
> with the timing of the seg fault). This is line 405 in the BEGIN
> statement for Bio::DB::Fasta
> --> @AnyDBM_File::ISA = qw(DB_File GDBM_File NDBM_File SDBM_File)
>
> I'm not sure what to do about those? Are you able to install GDBM_File,
> NDBM_File, or SDBM_File by themselves? They really should just be
> optional? I know I don't have NDBM_File, and I don't get an error.
> DB_File is part of the perl distribution, but can be updated via CPAN.
> Perhaps that would be good to do. It may just be that DB_file is broken
> and falling back to the other modules doesn¹t work because you don't have
> them. There is also the chance that the issue is really with Berkley_DB.
> DB_File accesses Berkley_DB and it wants version 1.x, but will work with
> 2.x and 3.x. You may need to even go as far as building a new Berkley_DB.
>
> Well that's enough to try for now.
>
> Thanks,
> Carson
>
>
> On 12-03-20 2:16 PM, "Thomas Hackl"<thomas.hackl at uni-wuerzburg.de> wrote:
>
>> Hi,
>>
>> I installed and ran the debug version of maker. Without -debug it
>> results in the expected segmentation fault. With -debug it paradoxically
>> finishes without the error.
>>
>> I screened the debug log anyway but could not find anything that helped
>> me to localize the problem. I attached a condensed version (sort | uniq)
>> of the log.
>> Any ideas are appreciated.
>>
>> Regards
>> Thomas
>>
>>
>>
>> Am 19.03.2012 17:24, schrieb Carson Holt:
>>> Ok. I finished the special debug version over the weekend. Run with
>>> -debug set as a command line flag. Capture and return the STDERR on
>>> failure. It will list all modules used, the versions, and when they are
>>> called so we can see what happens right before failure.
>>>
>>> http://yandell-lab.org/research/maker_debug.tgz
>>>
>>>
>>> Thanks,
>>> Carson
>>>
>>>
>>>
>>> On 12-03-14 1:57 AM, "Felix Bemm"<felix.bemm at uni-wuerzburg.de> wrote:
>>>
>>>> Hi,
>>>>
>>>> Thomas and I are colleagues and dealing with the same problem here. I
>>>> wrote some test code that forced the indexing and it work fine. We are
>>>> using most of the bioperl seqio and db modules in combination with
>>>> other
>>>> tools and don't experience the same problem there at the moment.
>>>>
>>>> Regards
>>>> Felix
>>>>
>>>> Am 13.03.2012 22:37, schrieb Barry Moore:
>>>>> You might also try a short perl script outside of MAKER to exercise
>>>>> Bio::DB::Fasta (which I believe is the module MAKER uses for Fasta
>>>>> indexing - correct Carson?).
>>>>>
>>>>> Something like this should work to force indexing:
>>>>>
>>>>> useBio::DB::Fasta;
>>>>> my $db = Bio::DB::Fasta->new('/path/to/fasta/files');
>>>>> my$seq=$db->seq($seqid, $start, $end);
>>>>>
>>>>> Point it at your fasta directory or file.
>>>>>
>>>>> B
>>>>>
>>>>> On Mar 13, 2012, at 3:41 AM, Thomas Hackl wrote:
>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> We reinstalled the packages you suggested
>>>>>>
>>>>>> forks is up to date (0.34).
>>>>>> forks::shared is up to date (0.34).
>>>>>> Inline::C is up to date (0.50).
>>>>>> Storable is up to date (2.30).
>>>>>>
>>>>>> as well as BioPerl.
>>>>>>
>>>>>> The problem is still the same.
>>>>>>
>>>>>> With MPI it terminates with this message:
>>>>>>
>>>>>> STATUS: Parsing control files...
>>>>>> STATUS: Processing and indexing input FASTA files...
>>>>>>
>>>>>>
>>>>>> ======================================================================
>>>>>> ==
>>>>>> =============
>>>>>>
>>>>>>
>>>>>> = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
>>>>>> = EXIT CODE: 11
>>>>>> = CLEANING UP REMAINING PROCESSES
>>>>>> = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
>>>>>>
>>>>>>
>>>>>> ======================================================================
>>>>>> ==
>>>>>> =============
>>>>>>
>>>>>> APPLICATION TERMINATED WITH THE EXIT STRING: Segmentation fault
>>>>>> (signal 11)
>>>>>>
>>>>>> which I presume is also the same problem.
>>>>>>
>>>>>> I am with you that the error is caused somewhere on the C level. If
>>>>>> the indexing step
>>>>>> is handled by non-maker modules exclusively, than the fact that the
>>>>>> 2.15 version works,
>>>>>> suggests that you are using different modules/methods in the current
>>>>>> releases?
>>>>>>
>>>>>> In any case, I think a version with verbose status messages might
>>>>>> help
>>>>>> to localize the
>>>>>> source of the problem.
>>>>>>
>>>>>> Regards
>>>>>> Thomas
>>>>>>
>>>>>> Am 07.03.2012 23:45, schrieb Carson Holt:
>>>>>>> There should be no new hardware requirement. But there is always a
>>>>>>> chance
>>>>>>> that there is an issue with one of the perl modules being used. I
>>>>>>> assume
>>>>>>> the failure is happening when using maker serially and you are not
>>>>>>> using
>>>>>>> MPI.
>>>>>>>
>>>>>>> Could you reinstall the following perl modules, and try again. If
>>>>>>> you
>>>>>>> are
>>>>>>> using CPAN, do a 'force install' to force it to reinstall.
>>>>>>>
>>>>>>> Modules:
>>>>>>> *Storable
>>>>>>> *Inline::C
>>>>>>> *forks
>>>>>>> *forks::shared
>>>>>>>
>>>>>>> Also try reinstalling the latest version of BioPerl.
>>>>>>>
>>>>>>> The fact that this is a seg fault suggests that something is
>>>>>>> happemning at
>>>>>>> the C level (just outside of Perl). Those area all the modules MAKER
>>>>>>> uses
>>>>>>> that will call back to the C level. BioPerl has a fasta indexing
>>>>>>> module
>>>>>>> that is also making calls outside of Perl, and the fact it fails at
>>>>>>> that
>>>>>>> point makes it a suspect.
>>>>>>>
>>>>>>> Let me know what happens. I can always generate an alternate MAKER
>>>>>>> executable for you to run with additional status messages that may
>>>>>>> help
>>>>>>> identify exactly which module is being called right before the
>>>>>>> failure.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Carson
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On 12-03-07 3:31 AM, "Thomas Hackl"<thomas.hackl at uni-wuerzburg.de
>>>>>>> <mailto:thomas.hackl at uni-wuerzburg.de>> wrote:
>>>>>>>
>>>>>>>> Hello,
>>>>>>>>
>>>>>>>> we want to use a current release of maker (2.22, 2.23) but the
>>>>>>>> program
>>>>>>>> terminates with a seg fault while processing the input FASTA files.
>>>>>>>> maker 2.15 , which we have been using for quite some time, runs
>>>>>>>> perfectly with identical data and setup.
>>>>>>>>
>>>>>>>> Please contact me if you need specifics on hardware, OS or anything
>>>>>>>> else.
>>>>>>>>
>>>>>>>> Best regards
>>>>>>>> Thomas
>>>>>>>>
>>>>>>>> --
>>>>>>>> Thomas Hackl
>>>>>>>> Julius-Maximilians-Universität
>>>>>>>> Department of Bioinformatics
>>>>>>>> 97074 Würzburg, Germany
>>>>>>>> Fon: +49 931 - 31 86883
>>>>>>>> Mail: thomas.hackl at uni-wuerzburg.de
>>>>>>>> <mailto:thomas.hackl at uni-wuerzburg.de>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> maker-devel mailing list
>>>>>>>> maker-devel at box290.bluehost.com
>>>>>>>> <mailto:maker-devel at box290.bluehost.com>
>>>>>>>>
>>>>>>>>
>>>>>>>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.
>>>>>>>> or
>>>>>>>> g
>>>>>> --
>>>>>> Thomas Hackl
>>>>>> Julius-Maximilians-Universität
>>>>>> Department of Bioinformatics
>>>>>> 97074 Würzburg, Germany
>>>>>> Fon: +49 931 - 31 86883
>>>>>> Mail: thomas.hackl at uni-wuerzburg.de
>>>>>> <mailto:thomas.hackl at uni-wuerzburg.de>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> maker-devel mailing list
>>>>>> maker-devel at box290.bluehost.com
>>>>>> <mailto:maker-devel at box290.bluehost.com>
>>>>>>
>>>>>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.or
>>>>>> g
>>>>> Barry Moore
>>>>> Research Scientist
>>>>> Dept. of Human Genetics
>>>>> University of Utah
>>>>> Salt Lake City, UT 84112
>>>>> --------------------------------------------
>>>>> (801) 585-3543
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> maker-devel mailing list
>>>>> maker-devel at box290.bluehost.com
>>>>>
>>>>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
>>>> _______________________________________________
>>>> maker-devel mailing list
>>>> maker-devel at box290.bluehost.com
>>>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
>> --
>> Thomas Hackl
>> Julius-Maximilians-Universität
>> Department of Bioinformatics
>> 97074 Würzburg, Germany
>> Fon: +49 931 - 31 86883
>> Mail: thomas.hackl at uni-wuerzburg.de
>>
>> _______________________________________________
>> maker-devel mailing list
>> maker-devel at box290.bluehost.com
>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
--
Thomas Hackl
Julius-Maximilians-Universität
Department of Bioinformatics
97074 Würzburg, Germany
Fon: +49 931 - 31 86883
Mail: thomas.hackl at uni-wuerzburg.de
More information about the maker-devel
mailing list