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

Metgrid process killed


New member
I'm trying to setup WRF for the first time and I got stuck when running metgrid.exe
The program runs for while and suddenly stops, showing only the message "Killed". The log file does not show any warning/error message (and it is too large to upload ~8.5Mb), and I have also checked that there is space available.
What could be causing this problem?
I'm using WRF 4.3.3 and WPS 4.3.1 versions. The libraries were installed using miniconda. Also, I'm using ERA5 as boundary data. Let me know if you need any other information.

Hi Laura,
Can you attach your namelist.wps file? Did you compile WPS for serial or parallel (dmpar) processing? Thanks!
I think I used the serial processing, but I'm not quite sure. Is there a way to verify this information in any of the output files? If that's possible, I can have a look tomorrow, because I don't have access to the cluster today.


  • namelist.wps
    882 bytes · Views: 3
Your domains (especially d02) are pretty large.
d01: 390 x 290
d02: 1201 x 741

The issue is likely that you can't run this with just a single processor. I suspect if you just tried to run metgrid.exe for domain 1, it would work. If that's the case, I believe you will need to recompile the WPS code for parallel computing (choose the "dmpar" option during configuration). You shouldn't need to re-run geogrid or ungrib. Just rerun metgrid, but do this by requesting multiple processors. The command will be something like:
mpiexec -np 16 ./real.exe
You may need to discuss this with a systems administrator to confirm the appropriate command for your machine.

If you haven't already, you will also need to compile WRF for parallel processing, as well. You will not be able to simulate with a domain of this size with a single processor.
Hi kwerner,
Thanks for the explanation!
I had previously compiled WRF for parallel processing.
I already have some met_em files for the 2nd domain, so I was wondering if it is really necessary to have all the simulation times for the 2nd domain as met_em files, since dom01 would provide the boundary conditions for the inner domain.
It is not necessary unless you are planning to use any type of nudging option or the sst_update option. If using any of those, you will need to process all times for all domains during WPS.