segmentation fault occurred when read initial condition

[Rocky_Wang]

Hello everyone,

I usually use the COAWST model to couple the ROMS and SWAN and simulate sediment transport. In recent days, I try to install COAWST model in a new server. I download the recent version of the COAWST and successfully run the UPWELLING case, the coupled Inlet_Test case.
However, when I try to run my real case, segmentation fault occurred when the model was reading the initial condition.This initial field file has been used many times and there has never been a problem. the segmentation fault as follow (also see the attachment Full_Simulation.log):

NLM: GET_STATE - Reading state initial conditions, 2017-12-01 00:00:00.00
(Grid 01, t = 0.0000, File: SD_grid2_ini.nc, Rec=0001, Index=1)
- free-surface
(Min = -2.69611686E-01 Max = 8.02609384E-01)
- vertically integrated u-momentum component
(Min = -2.85980076E-01 Max = 2.88503349E-01)
- vertically integrated v-momentum component
(Min = -4.65170592E-01 Max = 3.34708631E-01)
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
coawstM 000000000089F413 Unknown Unknown Unknown
libpthread-2.17.s 00007F8F6B74E6D0 Unknown Unknown Unknown
coawstM 00000000006CF98A nf_fread3d_mod_mp 156 nf_fread3d.f90
coawstM 000000000076185E get_state_ 685 get_state.f90
coawstM 00000000004244F8 initial_ 203 initial.f90
coawstM 0000000000423D9A ocean_control_mod 133 ocean_control.f90
coawstM 0000000000423294 MAIN__ 95 master.f90
coawstM 0000000000423062 Unknown Unknown Unknown
libc-2.17.so 00007F8F6B394445 __libc_start_main Unknown Unknown
coawstM 0000000000422F69 Unknown Unknown Unknown



I turn on the 'define ANA_INITIAL' cpp option to check if the model can run without the initial file. But I get anohter segmentation fault when reading the tidal foring file:

ANA_INITIAL - vertically integrated u-momentum component: ubar
(Grid = 01, Min = 0.00000000E+00 Max = 0.00000000E+00)
ANA_INITIAL - vertically integrated v-momentum component: vbar
(Grid = 01, Min = 0.00000000E+00 Max = 0.00000000E+00)
ANA_INITIAL - free-surface: zeta
(Grid = 01, Min = 0.00000000E+00 Max = 0.00000000E+00)
ANA_INITIAL - u-momentum component: u
(Grid = 01, Min = 0.00000000E+00 Max = 0.00000000E+00)
ANA_INITIAL - v-momentum component: v
(Grid = 01, Min = 0.00000000E+00 Max = 0.00000000E+00)
ANA_INITIAL - potential temperature: temp
(Grid = 01, Min = 1.00000000E+01 Max = 1.00000000E+01)
ANA_INITIAL - salinity: salt
(Grid = 01, Min = 3.00000000E+01 Max = 3.00000000E+01)
GET_NGFLD - tidal period
(Grid = 01, Min = 4.30820441E+04 Max = 9.67260855E+04)
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
coawstM 0000000000868E83 Unknown Unknown Unknown
libpthread-2.17.s 00007F27284B66D0 Unknown Unknown Unknown
coawstM 000000000070D529 distribute_mod_mp 4753 distribute.f90
coawstM 00000000006B214B nf_fread3d_mod_mp 245 nf_fread3d.f90
coawstM 0000000000640320 get_2dfld_ 213 get_2dfld.f90
coawstM 00000000004353EA get_idata_ 86 get_idata.f90
coawstM 000000000042457C initial_ 305 initial.f90
coawstM 0000000000423A1A ocean_control_mod 133 ocean_control.f90
coawstM 0000000000422F14 MAIN__ 95 master.f90
coawstM 0000000000422CE2 Unknown Unknown Unknown
libc-2.17.so 00007F27280FC445 __libc_start_main Unknown Unknown
coawstM 0000000000422BE9 Unknown Unknown Unknown


I have no idea about that, Please give me some suggestions and help, thanks very much!

Rocky Wang

[jcwarner]

not sure. but both inlet test and upwelling use ana_initial. So how about you try something that needs an initial netcdf file. You could try our Sandy case, set it up to just use roms only (undef swan and undef wrf). That case uses a netcdf initial file. lets see if that works.
-j

[Rocky_Wang]

Hi John

Thanks for your quick reply, It's a good idea to test the Sandy case with initial file. I run the sandy case with three data files : Sandy_roms_grid.nc, Sandy_ini.nc and tide_forc_Sandy.nc, both of them were from COAWST/Projects/Sandy and just the ROMS model was turned on. At this condition, the model can run successfully. The log file is Sandy+initial+tide.log in the attachment.

Then I use my matlab code (from COAWST/tools) to create a same initial file （Sandy_ini_wch.nc） for the Sandy, to check if the there is somthing wrong with my matlab code. and also create a tidal foring file （Sandy_tide_tpxo8.nc）using TPXO2ROMS_v4pt0.m. The model can also run successfully, The log file is Sandy+New_initial+New_tide.log in the attachment.

It seems like both the COAWST installation and Matlab code are good.The I use the same Matlab code to create an new initial file （Bysea_New_ini.nc）and tide file （SD_grid2_tide_tpxo8.nc）for my case, and use same model configuration with Sandy . The most confusing thing is, the segmentation fault still occurred when read initial condition and the model seems like can't read 3-D variables. The log file is Bysea+New_initial+New_tide.log in the attachment.

This is really strange. My model grid has been used for a long time, so I don’t think it’s the problem. If you have time, could please check my simulation records to see if there are some clues.

Thank you！

Rocky Wang

[jcwarner]

suggest you do this.
recompile with DEBUG=on, that will give a coawstG
run it with just 1 proc, this will make the error more easy to see.