[cam-users] CCM3 segmentation fault on Sun solved

[Curtis Covey]

Thanks, Cathy!  With the C-shell command "limit stacksize
unlimited,"  the model now runs a full day as expected.  The
previous stacksize limit was 8192 kbyes.

- Curt Covey

Cathryn Meyer wrote:

>I have been having the same issues as you recently, and am usually just
>as clueless regarding other people's suggestions.  So to start, you want
>to try setting 'limit stacksize unlimited'.  All you have to do is type
>that on the command line in your terminal window, if you are in tc-shell
>or c-shell.  In order to get into tcshell, just type tcsh on your
>command line.  You can also type 'setenv MPSTKZ 40000000' and that might
>help your memory issues as well.  If you just type limit on the command
>line, it will tell you what your limits are.  Any of those terms
>(stacksize, datasize, etc) can be set to unlimited using the terminology
>above.  
>
>However none of these helped to solve my problem, and I still get
>segmentation faults and can't run the model, and I have not solved this
>problem yet.  But most people are able to fix their segmentation faults
>with these few things.
>
>Cathy
>
>Quoting Curtis Covey <covey1 at llnl.gov>:
>
>  
>
>>Dear CAM Users Group,
>>
>>I am trying to run CCM3.6.6 on the Sun server tethys.llnl.gov,
>>under SunOS version 5.8.  The model compiles and links without
>>apparent problems after minimal changes to build.noncray.csh and
>>Makefile from the standard code distribution.  Upon execution,
>>however, the model quickly terminates with a segmentation fault
>>and core dump.  The tail end of the atm.log file is:
>>
>> ---------------------------------------
>> ---------------------------------------
>> creating lsm time-constant history file
>> lsm time constant dataset fields:
>> no name     units    description
>>  1 SUB_LAT  degrees_ subgrid vector latitude
>>
>>In the archives of Users Group correspondence,  Jim Rosinski
>>suggests another user's seg fault is the result of exceeding
>>some system memory limit, which could possibly be fixed with
>>things like "limit stacksize unlimited", "limit memoryuse
>>unlimited", etc. (in csh).  Also, Kalyana Chadalavada finds that
>>setting large limits (~1 GB) for stack size, data seg size, max
>>mem and max locked mem solved his seg fault problem.
>>
>>This all sounds reasonable, but I am sufficiently inexperienced
>>that I don't know how to implement these suggestions.  Any
>>advice as to the best way to proceed?
>>
>>- Curt Covey
>>
>>_______________________________________________
>>cam-users mailing list
>>cam-users at ucar.edu
>>http://mailman.ucar.edu/mailman/listinfo/cam-users
>>
>>    
>>
>
>
>  
>