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

MPAS segmentation error when running 30-km resolution though successful in 120-km.

f930139

New member
Hi everyone
I am running MPAS-A 7.0 with ungrib ERA5 data (30-km meshes), but the ./atmosphere_model failed when outputing output.nc and diag.nc. slurm script shows 70220 lines of " Flerchinger USEd in NEW version. Iterations= 10" and then Segmentation fault.
There's no error when ./init_atmosphere_model, including geog data, met data and sst update data.
I also use the same ungrib ERA5 data to run 120-km meshes, and it works well (empty slurm script).

I guess the problem lies in inital data, but I cannot find where the problem is.
Did anyone face with the same problem before? Is it just simply related to the MPAS version, or the 30-km meshes are problematic?

30-km meshes are downloaded from MPAS-Atmosphere mesh downloads
 

Attachments

  • namelist.init_atmosphere(init_met).txt
    1.4 KB · Views: 0
  • namelist.init_atmosphere(init_sst).txt
    1.4 KB · Views: 0
  • namelist.init_atmosphere(static).txt
    1.4 KB · Views: 0
  • streams.init_atmosphere(init_met).txt
    952 bytes · Views: 0
  • streams.init_atmosphere(init_sst).txt
    952 bytes · Views: 0
  • streams.init_atmosphere(static).txt
    952 bytes · Views: 0
  • namelist.atmosphere.txt
    1.8 KB · Views: 2
  • streams.atmosphere.txt
    1.6 KB · Views: 1
  • slurm-5799294.out.txt
    706.9 KB · Views: 1
Would you please upload your ERA5 data of me to take a look? Did you use the same ERA5 data to run 120-km mesh successfully? Thanks.
 
Hi everyone
I am running MPAS-A 7.0 with ungrib ERA5 data (30-km meshes), but the ./atmosphere_model failed when outputing output.nc and diag.nc. slurm script shows 70220 lines of " Flerchinger USEd in NEW version. Iterations= 10" and then Segmentation fault.
There's no error when ./init_atmosphere_model, including geog data, met data and sst update data.
I also use the same ungrib ERA5 data to run 120-km meshes, and it works well (empty slurm script).

I guess the problem lies in inital data, but I cannot find where the problem is.
Did anyone face with the same problem before? Is it just simply related to the MPAS version, or the 30-km meshes are problematic?

30-km meshes are downloaded from MPAS-Atmosphere mesh downloads
Hi
I have had the same issue.
Can you tell me what is the system configuration that you are working on? like how many cores are you working on and also the memory (RAM).
 
Hi
I have had the same issue.
Can you tell me what is the system configuration that you are working on? like how many cores are you working on and also the memory (RAM).
I solved this segmentation problem by changing dt from 720.0 to 180.0 seconds. I'm not sure if this can solve your problem as well.
I am working on 16 nodes and 48 tasks each node.
 
Top