[gmx-users] Segmentation Fault (Address not mapped)

darrellk at ece.ubc.ca darrellk at ece.ubc.ca
Tue Jul 14 01:13:12 CEST 2009


Hi Justin,
I was experiencing the problem before someone suggested using editconf so
I do not think the problem is being caused by editconf. But anyway here
is my editconf command. Let me know if you a source of error in this
command line.

editconf -f graphene.gro -n index.ndx -o graphene_ec.gro

I did not want to add in additional space between the solvent and the box
as I saw no reason for doing so. And hence that is why I originally did
not use editconf.

My box dimensions are 38nm x 38nm x 38nm. I used cutoffs of 2 nm & 5 nm
for my system so ensure the cutoff occured at a distance where the
potentials were stabalized (not changing). I guess I could use shorter
cutoffs such as 1.5 nm & 2 nm and this may decrease my computation time.
I also thought that I needed to use larger cut-offs since I am dealing
in the gas phase and there is greater ditance between the atoms in my
simulation than in liquid-based simulations.

In the .log files, I did not see any LINCS warnings or neighborlist
errors.

The fault appears to be occurring sometime between 6 ps & 8 ps (between
time step 3000 & time step 4000) since, when I view the .trr file in
ngmx, I only see frames for 2 ps, 4 ps, and 6 ps. Note that I write
frames every 1000 steps and my time step is 0.002 ps.

I ran gmxcheck on the .trr file and was presented with the following
output:
*********************************************
Checking file mdtraj.trr
trn version: GMX_trn_file (single precision)
Reading frame       0 time    0.000
# Atoms  10482
Last frame          5 time    1.000


Item        #frames Timestep (ps)
Step             6    0.2
Time             6    0.2
Lambda           6    0.2
Coords           6    0.2
Velocities       6    0.2
Forces           0
Box              6    0.2
*********************************************

I am currently running another simulation with a reduced value for
nstxout. Hopefully this will provide me with more information in the .lg
files so that I may locate the source of my problem.

I will report what I find once my simulation completes. In the meantime,
if any of the above information provides you with clues as to the source
of my problem, please let me know.

Thank you very much.

Darrell

>Date: Mon, 13 Jul 2009 15:37:15 -0400
>From: "Justin A. Lemkul" <jalemkul at vt.edu>
>Subject: Re: [gmx-users] Segmentation Fault (Address not mapped)
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Message-ID: <4A5B8CEB.4020609 at vt.edu>
>Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
>
>
>darrellk at ece.ubc.ca wrote:
>> Hi Mark,
>> I used editconf on my .gro file with zero space between my solvent and
>> the box and the resulting box had the exact same dimension as the
>> initial box. I also performed a number of simulation runs with different
>
>If you're using editconf to define zero space, what's the point?  I only ask
>because it is a potential source of error if you think you're adding zero space,
>but something else might be going on.  Maybe you can post your editconf command
>line.
>
>What are your box dimensions?  Are cut-off lengths of 2.0 and 5.0 nm appropriate
>for your system?  How did you determine that these cut-off's should be used?
>
>> mdp parameters hoping this would provide me some indication of the cause
>> of the fault but to no avail. I looked through the log files, error
>> files, and output files and could not find any output to help me
>> identify the source of my error.
>>
>
>It is very odd that Gromacs isn't report anything at all.  No LINCS warnings?
>No neighborlist errors?  These would be in the .log file.
>
>> Could you please let me know how I can look at my structure at each point
>> as you indicate below as I do not see any files output that provide me
>> to do so? I tried to look at the .trr file but when I try to load it
>> into VMD, it causes an error. I am assuming this error is caused because
>> the .trr file did not complete correctly due to the segmentation fault.
>> Please advise.
>>
>
>How early is the segmentation fault occurring?  I have found it useful sometimes
>to set nstxout (or nstxtcout) = 1 to try to catch the first few frames if the
>explosion is occurring early.  In any case, gmxcheck will help determine how
>many frames are present, as well as the integrity of the file (broken frames, etc).
>
>-Justin
>
>> Thanks.
>>
>> Darrell
>>
>>> Date: Tue, 07 Jul 2009 09:19:42 +1000
>>> From: Mark Abraham <Mark.Abraham at anu.edu.au>
>>> Subject: Re: [gmx-users] Segmentation Fault (Address not mapped)
>>> To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>>> Message-ID: <4A52868E.6010807 at anu.edu.au>
>>> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>>>
>>> darrellk at ece.ubc.ca wrote:
>>>> Hi Mark,
>>>> I added the energy group exclusions as indicated in your previous
>>>> response but am still experiencing the same problem. I looked at the
>>>> .log files and see that in one log file it tells me that my box is
>>>> exploding. However, I do not have many molecules in my simulation and
>>>> therefore do not think that it is possible that my box is exploding from
>>>> pressure.
>>> Sure, but if there's something malformed with your model physics or
>>> starting configuration, then large forces can make anything explode.
>>>
>>> Look at your structures at each point and see where things start to go
>>> wrong. Make sure you've used editconf on your starting structure to
>>> provide the right box dimensions.
>>>
>>> Mark
>>>
>>>> Maybe if I re-state my simulation it will help you in providing me
>>>> direction on what might be causing the problem. My simulation consists
>>>> of a graphene lattice with a layer of ammonia molecules above it. The
>>>> box is very large and there is lots of empty space in the box. So I am a
>>>> little confused as to how the box could be exploding.
>>>>
>>>> Thanks again in advance for your help.
>>>>
>>>> Darrell Koskinen
>>>>
>>>>> Date: Fri, 03 Jul 2009 11:41:45 +1000
>>>>> From: Mark Abraham <Mark.Abraham at anu.edu.au>
>>>>> Subject: Re: [gmx-users] Segmentation Fault (Address not mapped)
>>>>> To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>>>>> Message-ID: <4A4D61D9.6080700 at anu.edu.au>
>>>>> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>>>>>
>>>>> darrellk at ece.ubc.ca wrote:
>>>>>> Dear GROMACS Gurus,
>>>>>> I am experiencing a segmentation fault when mdrun executes. My simulation
>>>>>> has a graphene lattice with an array (layer) of ammonia molecules above
>>>>>> it. The box is three times the width of the graphene lattice, three
>>>>>> times the length of the graphene lattice, and three times the height
>>>>>> between the graphene lattice and the ammonia molecules. I am including
>>>>>> the mdp file and the error message.
>>>>> Probably your system is exploding when integration fails with excessive
>>>>> forces. You should look at the bottom of stdout, stderr, *and* the .log
>>>>> file to diagnose. The error message you give below is merely the
>>>>> diagnostic trace from the MPI library, and it not useful for finding out
>>>>> what GROMACS thinks the problem might be. Further advice below.
>>>>>
>>>>>> ***************************************************************************
>>>>>> .mdp file
>>>>>> title           =FWS
>>>>>> ;warnings       =10
>>>>>> cpp             =cpp
>>>>>> ;define         =-DPOSRES
>>>>>> ;constraints    =all-bonds
>>>>>> integrator      =md
>>>>>> dt              =0.002 ; ps
>>>>>> nsteps          =100000
>>>>>> nstcomm         =1000
>>>>>> nstxout         =1000
>>>>>> ;nstvout                =1000
>>>>>> nstfout         =0
>>>>>> nstlog          =1000
>>>>>> nstenergy       =1000
>>>>>> nstlist         =1000
>>>>>> ns_type         =grid
>>>>>> rlist           =2.0
>>>>>> coulombtype     =PME
>>>>>> rcoulomb        =2.0
>>>>>> vdwtype         =cut-off
>>>>>> rvdw            =5.0
>>>>>> fourierspacing  =0.12
>>>>>> fourier_nx      =0
>>>>>> fourier_ny      =0
>>>>>> fourier_nz      =0
>>>>>> pme_order       =4
>>>>>> ewald_rtol      =1e-5
>>>>>> optimize_fft    =yes
>>>>>>
>>>>>> ; This section added in to freeze hydrogen atoms at edge of graphene
>>>>>> lattice to prevent movement of lattice
>>>>>> ;energygrp_excl = Edge Edge Edge Grph Grph Grph
>>>>>> freezegrps      = Edge Grph ; Hydrogen atoms in graphene lattice are
>>>>>> associated with the residue Edge
>>>>> See comments in 7.3.24 of manual. You need the energy group exclusions.
>>>>>
>>>>> Mark
>>>>>
>>>>>> freezedim       = Y Y Y Y Y Y; Freeze hydrogen atoms in all directions
>>>>>>
>>>>>> ;Tcoupl         =berendsen
>>>>>> ;tau_t          =0.1    0.1
>>>>>> ;tc-grps                =protein non-protein
>>>>>> ;ref_t = 300 300
>>>>>>
>>>>>> ;Pcoupl = parrinello-rahman
>>>>>> ;tau_p = 0.5
>>>>>> ;compressibility = 4.5e-5
>>>>>> ;ref_p = 1.0
>>>>>>
>>>>>> ;gen_vel = yes
>>>>>> ;gen_temp = 300.0
>>>>>> ;gen_seed = 173529
>>>>>> ***************************************************************************
>>>>>>
>>>>>> ***************************************************************************
>>>>>> ERROR IN OUTPUT FILE
>>>>>> [node16:25758] *** Process received signal ***
>>>>>> [node16:25758] Signal: Segmentation fault (11)
>>>>>> [node16:25758] Signal code: Address not mapped (1)
>>>>>> [node16:25758] Failing at address: 0xfffffffe1233e230
>>>>>> [node16:25758] [ 0] /lib64/libpthread.so.0 [0x3834a0de80]
>>>>>> [node16:25758] [ 1] /usr/lib64/libmd_mpi.so.4(pme_calc_pidx+0xd6)
>>>>>> [0x2ba295dd0606]
>>>>>> [node16:25758] [ 2] /usr/lib64/libmd_mpi.so.4(do_pme+0x808)
>>>>>> [0x2ba295dd4058]
>>>>>> [node16:25758] [ 3] /usr/lib64/libmd_mpi.so.4(force+0x8de)
>>>>>> [0x2ba295dba5be]
>>>>>> [node16:25758] [ 4] /usr/lib64/libmd_mpi.so.4(do_force+0x5ef)
>>>>>> [0x2ba295ddeaff]
>>>>>> [node16:25758] [ 5] mdrun_mpi(do_md+0xe23) [0x411193]
>>>>>> [node16:25758] [ 6] mdrun_mpi(mdrunner+0xd40) [0x4142f0]
>>>>>> [node16:25758] [ 7] mdrun_mpi(main+0x239) [0x4146f9]
>>>>>> [node16:25758] [ 8] /lib64/libc.so.6(__libc_start_main+0xf4)
>>>>>> [0x3833e1d8b4]
>>>>>> [node16:25758] [ 9] mdrun_mpi [0x40429a]
>>>>>> [node16:25758] *** End of error message ***
>>>>>> mpirun noticed that job rank 7 with PID 25758 on node node16 exited on
>>>>>> signal 11 (Segmentation fault).
>>>>>> 7 processes killed (possibly by Open MPI)
>>>>>> ***************************************************************************
>>>>>>
>>>>>> Could you please let me know what you think may be causing the fault?
>>>>>>
>>>>>> Much thanks in advance.
>>>>>>
>>>>>> Darrell Koskinen
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
>
>--
>========================================
>
>Justin A. Lemkul
>Ph.D. Candidate
>ICTAS Doctoral Scholar
>Department of Biochemistry
>Virginia Tech
>Blacksburg, VA
>jalemkul[at]vt.edu | (540) 231-9080
>http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin



More information about the gromacs.org_gmx-users mailing list