seg fault and error when using "inithist='none'" in namelist

[lmurphy@atmos_umd_edu]

Hi all,

I am running the CAM3-SOM on bluevista. It runs fine for the first 12 months but then when I restart I get a segmentation fault right after the model writes the clm initial conditions file. Jerry Olson said that there is some sort of compiler problem on bluevista and I followed his suggestion to add "inithist = 'none'" in my namelist but it's still not running. Now it quits before it initializes the land model. After putting together the Master field list it says that it cannot find the variable LANDM_COSLAT, a variable that I find in the initial conditions file and not the history files. The end of my output file is pasted below. Thanks in advance for any suggestions.

Lisa
******************************************************************************************************
471 TSICERAD&IC K
472 TBOT&IC K
ENDRUN:GET_MASTERLIST_INDX: attemping to output field LANDM_COSLAT not on master list

Traceback:
Offset 0x000000bc in procedure __abortutils_NMOD_endrun, near line 45 in file /homebv/mahowald/andrewscode/models/atm/cam/src/contr
ol/abortutils.F90
Offset 0x000002f4 in procedure __history_NMOD_get_masterlist_indx, near line 6431 in file /homebv/mahowald/andrewscode/models/atm/c
am/src/control/history.F90
Offset 0x000001e4 in procedure __history_NMOD_bld_htapefld_indices, near line 6581 in file /homebv/mahowald/andrewscode/models/atm/
cam/src/control/history.F90
Offset 0x000073ac in procedure __history_NMOD_bldfld, near line 4958 in file /homebv/mahowald/andrewscode/models/atm/cam/src/contro
l/history.F90
Offset 0x00000e38 in procedure cam, near line 191 in file /homebv/mahowald/andrewscode/models/atm/cam/src/control/cam.F90
--- End of call chain ---
ERROR: 0031-250 task 6: Terminated
ERROR: 0031-250 task 2: Terminated
ERROR: 0031-250 task 4: Terminated
ERROR: 0031-250 task 0: Terminated
ERROR: 0031-250 task 3: Terminated
ERROR: 0031-250 task 7: Terminated
ERROR: 0031-250 task 1: Terminated
ERROR: 0031-250 task 5: Terminated
Job /usr/local/lsf/7.0/aix5-64/bin/poejob -euidevice sn_all -euilib us /ptmp/lisa/b42.lmaxp.nodust/bld/cam

TID HOST_NAME COMMAND_LINE STATUS TERMINATION_TIME
===== ========== ================ ======================= ===================
00006 bv0805en.u /ptmp/lisa/b42.l Exit (143) 08/04/2008 12:01:18
00007 bv0805en.u /ptmp/lisa/b42.l Exit (143) 08/04/2008 12:01:18
00004 bv1004en.u /ptmp/lisa/b42.l Exit (143) 08/04/2008 12:01:18
00005 bv1004en.u /ptmp/lisa/b42.l Exit (143) 08/04/2008 12:01:18
00002 bv0801en.u /ptmp/lisa/b42.l Exit (143) 08/04/2008 12:01:18

 

[olson]

First, a stupid question: just want to make sure you didn't use the double quotes when
including inithist = 'none' in your namelist

That out of the way, did you use inithist = 'none' on your restart run only or did you start all
over again by including the inithist = 'none' in your original initial run script? I ask since the
model can get confused and exhibit unpredictable behavior if the namelist is altered on a restart.
You need to include the inithist line in your initial run as well as restarts.

Also, it still isn't clear to me if you're running into the SOM compiler bug which I'm not
as familiar with. I think you need to compile the model with both the -smp and -spmd compiler
flags to avoid the SOM problem, if I remember correctly. Are both flags being used in your
configure?

 

[lmurphy@atmos_umd_edu]

Thanks for replying so fast! No, I didn't use double quotes but that was a good question. :) However, I added the inithist option to the restart namelist. So that's a very good point and I didn't think even think about that. I'll recompile and start over. Thanks!

Lisa

 

[lmurphy@atmos_umd_edu]

Quick question....in my build script I had:

$cfgdir/configure
-fflags=-WF,-DDUST,-DBGC,-DPROGSSLT,-DDUSTFEEDBACK -nadv 11
-ocn som
-dyn $dycore
-res $resolution
-usr_src ${usr_src},${som_src}

Should I add the compiling options after "-ocn som"?

thanks,
Lisa

 

[lmurphy@atmos_umd_edu]

Sorry stupid question. I see it in the configure file. Nevermind!