Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

segmentation fault


Dear forum member,
I started a simulation with the 2 last domains as LES. Unfortunately, after approximately 30 minutes, the simulation stops. I tested various settings (physics, radiation model) but was not successful. I perform the simulation on a cluster so that memory problem should not be a problem. I attach the namelist file and the error files. I would be very grateful for any help. I am using SRTM data for the DEM.


  • namelist.input
    7.1 KB · Views: 6
  • rsl.error.0000
    96.2 KB · Views: 1
There is no error massage in the rsl file you posted. Can you find any error messages in other rsl files that show reasons for the crash?
Thanks for your answer, unfortunately no, I only see in the command line that it crashed because of segmentation error, that s the reason why I have no idea where the problem comes everything ok with the setting? Or may it for some reason becomes unstable?
epssm =2.8, why is this value so hi @kermarrec

in the /run folder were the error files are run these commands and see if they come up with anything.

grep -i FATAL rsl.*

grep -i error rsl.*

grep -i SIGSEGV rsl.*

grep -i cfl rsl.*
I chose a high value because I have a mountainous terrain, that was just a try to see if it will work.
The output did not give anything but now I think that the problem was most probably coming from a time limitation. Writing the file
auxhist7_outname = "myfile71_d<domain>_<date>",
auxhist7_interval = 0, 0, 0, 0 , 1, !not needed for 3 coarse domains d1-3
seems extremely time-consuming. Is there a way to speed it up? thanks for your feedback
Can you modify namelist options as below, then try again:

time_step = 12,
time_step_fract_num = 0,
time_step_fract_den = 1,
cu_physics = 0, 0, 0, 0, 0,
bl_pbl_physics = 0, 0, 0, 0, 0,
sf_sfclay_physics = 1, 1, 1, 1, 1
diff_opt = 2, 2, 2, 2, 2
km_opt = 5, 5, 5, 5, 5
nio_tasks_per_group = 0,
nio_groups = 1,
epssm = 0.9

Please let me know whether this case can work with the above options.
If I understand correctly, it means I would not go for LES in the last 2 domains? I would like to have an output every 1-2 seconds to be able to have reliable spectra, is it impossible or computational "too" demanding?