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

atmosphere_model - bad termination of one of your application processes

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.



I am running atmosphere_model (the latest version available) and it is terminating suddenly for some reason. I get "your application terminated with the exit string: killed (signal 9). I will attach the log*out, the namelist file, and the error msg I see on screen. The log file didn't indicate anything obvious. Maybe you can help by seeing the files I am attaching.



  • error.msg.steve.txt
    531 bytes · Views: 73
  • namelist.atmosphere.steve.txt
    1.8 KB · Views: 80
  • log.atmosphere.0000.out.steve.txt
    8.3 KB · Views: 72
Your namelist contains an error. Between the &nhyd_model and the &damping sections there is an extra config_run_duration keyword probably put there by mistake; however, I don't think this is enough to crash the model, but its worth cleaning up.

Otherwise, if you haven't already done so, it might be worth cleaning and recompiling with DEBUG=true to generate a stack trace, which will hopefully give us a line number at the point of failure.
Hi - thanks for your quick response. I cleaned up the namelist file and also recompiled with the DEBUG=true option set. However, it is still stopping, after about 10 seconds with the "killed - signal 9" error message. I am running in a virtual machine, with 4GB of RAM. Since I have been attempting a global simulation I'm thinking it is running out of memory when it starts to allocate arrays. I would like to instead now set up a "regional" simulation and will check the User's notes and the tutorial notes for help on that. Hopefully that is the issue (lack of memory) but won't know until I can set up this new test.
Another option would be to do a test run using a smaller mesh if you have not already done so. For instance the 480-km mesh (2562 horizontal grid cells) which is 16x smaller than the 120 km mesh (40962 horizontal grid cells). That way, you can determine if your issue is related to memory or not.
I agree that the issue may be a memory limitation. A rough approximation for the memory required to run a simulation on a mesh with N cells is N * 0.175 MB, which suggests around 7.2 GB of memory would be required for the 120-km mesh with 40962 cells. There is more discussion on computational requirements in this thread that may be useful.
That was it - the smaller mesh (2562 grid cells) worked just fine. Thanks so much for the suggestion. And the rule-of-thumb for #cells and the required RAM is really useful. After I set things up so I can visualize the output I'm still interested in testing out a regional domain.