mpirun has exited

[Zhi Mao]

Hello,I want to run WRFV4,but there are errors after I run the "real" program.I am sure I follow the step:Build WRF—Build WPS—Run WPS—Run WRF,all the steps are ok until I execute "real.exe".I use Ubuntu 18.04,intel2019 compiler,openmpi—1.6.5(maybe it is the problem).The following is the screenshot of the result of "mpirun -np 1 real.exe",my bashrc file and rsl.error.0000.Look forward to you reply.I've been having this problem for days.Thank you very much!

 

[Ming Chen]

The screen shot is hard to read. Please attach your rsl file and namelist.input for me to take a look.

 

[Zhi Mao]

Because the format of "rsl.error.0000" and "rsl.out.0000" is not supported, I convert them to pdf.The namelist.iuput keep the original format.Thank you!

 

[adildilawar]

Dear Zhi Mao,
As far I know and i checked your namelist.input. I think the problem with your date for domains as below:
start_year = 2016, 2000, 2000,
start_month = 06, 01, 01,
start_day = 15, 24, 24,
start_hour = 00, 12, 12,
end_year = 2016, 2000, 2000,
end_month = 06, 01, 01,
end_day = 16, 25, 25,
end_hour = 00, 12, 12,

Please try to adjust it and re-run ungrib and so on. Lets see problem is same. Otherwise wait for the answer from the professionals here.

Best,
adil

 

[Ming Chen]

This case is run on a single domain, so the time settings for D02 and D03 don't matter.

Since it crashed immediately after starting due to segmentation fault, I am suspicious this is a memory issue.

I siuppose this code was compiled in dmpar mode (please let me know if I am wrong). Can you run with more processors, for example, mpirun -np 8 ./real.exe? Using more processors will give you more memory to run real.exe.

If you still cannot run the case with more processors, please provide the following information:
(1) Version of WRF
(2) Input data to drive the model
(3) Your namelist.wps

 

[Zhi Mao]

Adil,thank you for your help!
Dear Ming,my version of WRF is WRF Version 4.0(released this on 9 Jun 2018).When I was building WRF,I chose option 15(damper) after executed "./configure".When I was building WPS,I chose option 19(Linux x86_64,Inter compiler,damper) after executed "./configure".I just browsed "https://github.com/wrf-model/WRF/releases?after=v4.0.1",I found "Support for OpenMPI version 1 is removed" from the software section of WRF Version 4.0.I'm so bad to see the note because I was wrong at the first time(I used openmpi-1.6.5.),but I didn't know it.Maybe I should rebuild the library using mpich-3.0.4?

 

[Zhi Mao]

Dear Ming,
I uninstall openmpi-1.6.5,install mpich-3.0.4,and re-compile WRF & WPS.I don't change my namelist.wps and namelist.input.As you have said,mpirun -np 8 ./real.exe is OK,and all rsl.error files show "SUCCESS".Then I run "mpirun -np 8 ./wrf.exe",I got wrfout file,but my screen show:
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= EXIT CODE: 174
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
is that OK?

And my rsl.error.0000 shows:
taskid: 0 hostname: zhou
module_io_quilt_old.F 2931 F
Quilting with 1 groups of 0 I/O tasks.
Ntasks in X 2 , ntasks in Y 4
WRF V4.0 MODEL
*************************************
Parent domain
ids,ide,jds,jde 1 220 1 200
ims,ime,jms,jme -4 117 -4 57
ips,ipe,jps,jpe 1 110 1 50
*************************************
DYNAMICS OPTION: Eulerian Mass Coordinate
alloc_space_field: domain 1 , 149048308 bytes allocated
med_initialdata_input: calling input_input
Input data is acceptable to use: wrfinput_d01
Timing for processing wrfinput file (stream 0) for domain 1: 1.07701 elapsed seconds
Max map factor in domain 1 = 1.03. Scale the dt in the model accordingly.
INPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND 33 CATEGORIES 2 SEASONS WATER CATEGORY = 17 SNOW CATEGORY = 15
INITIALIZE THREE Noah LSM RELATED TABLES
Skipping over LUTYPE = USGS
LANDUSE TYPE = MODIFIED_IGBP_MODIS_NOAH FOUND 20 CATEGORIES
INPUT SOIL TEXTURE CLASSIFICATION = STAS
SOIL TEXTURE CLASSIFICATION = STAS FOUND 19 CATEGORIES
Timing for Writing wrfout_d01_2016-06-15_00:00:00 for domain 1: 2.77289 elapsed seconds
d01 2016-06-15_00:00:00 Input data is acceptable to use: wrfbdy_d01
Timing for processing lateral boundary for domain 1: 0.16877 elapsed seconds
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
wrf.exe 0000000002FCF8D3 for__signal_handl Unknown Unknown
libpthread-2.27.s 00007FDC33422890 Unknown Unknown Unknown
wrf.exe 000000000132C250 Unknown Unknown Unknown
wrf.exe 0000000001203400 Unknown Unknown Unknown
wrf.exe 0000000000555F7B Unknown Unknown Unknown
wrf.exe 000000000040F3B1 Unknown Unknown Unknown
wrf.exe 000000000040F36F Unknown Unknown Unknown
wrf.exe 000000000040F302 Unknown Unknown Unknown
libc-2.27.so 00007FDC32CA2B97 __libc_start_main Unknown Unknown
wrf.exe 000000000040F1EA Unknown Unknown Unknown
Does that mean I succeeded?

 

[Ming Chen]

This case failed at the right beginning. Your namelist.input looks fine.
I am still suspicious that this is a memory issue. If you are using Linux, check to see if there is any restriction on your memory use:

ulimit -a

If the memory is limited, you may need to increase it with "ulimit -m value" (value in kbytes or mbytes),
or ask the system administrator to do it.

 

[Zhi Mao]

Thank you for your help!I found the problem from FAQ,it was a segmentation fault!I typed "ulimit -s unlimited",changed time_step=3*dx,and added smooth_cg_topo = .true. in the &domains section.The wrfout file is OK.

 

[Zhi Mao]

Dear Ming,
I am doing my homework about urban canopy layer（UCL）.I want to select a typical urban area,and estimate the UCL height based on available data（actually I have no any data）,but the content about urban canopy model（UCM） is very limit in the WRFUsersGuide.Do you know how to do that?

 

[Ming Chen]

Hi, Zhi Mao,
Please see the attached document that describes urban canopy module. For your question of UCL height, my understanding is that it is specified based on types of urban surface. At present the model has no capability to describe sub-grid scale urban surface features.