Hello,
I am trying to run WRFv 4.5.2 on Derecho (using the pre-compiled version on Derecho), and have been running into issues during my second restart run. The model runs quickly for the initial run, and for the first restart run. However, on the second restart run, things slow down a lot, and it seems to be behaving oddly (shown in the rsl files attached below, for two different examples). The issue I see in the rsl files is when it goes to write out the first auxiliary output file (I am outputting some variables hourly, for analysis), it seemingly tries to write out the same file over and over again.
This must be why it is running slow, it is stalling on that one file. However, I cannot figure out what could be the issue here, I have toggled adaptive timestep on and off, to see if that was the issue, and the issue still persists. I also tried upping to 256 processors (originally using 128 processors) before beginning the second restart run, which did not make a difference, and the same issue persisted. There are no CFL violations, either.
I have attached namelists for two different WRF runs (different members of my ensemble, all having this issue), the corresponding rsl files, and the aux_var files. I also attached a screen capture of the rsl file during the first restart run (when the model was running well) for reference. Do you have any suggestions on why this is happening/see any issues? I have used this setup successfully for other simulations. Thank you in advance!
Here is the path to my 'ens3' and 'ens4' simulations on Derecho, if it helps:
/glade/derecho/scratch/asinning/ivan/ens3_ivan_present_wdm7rerun/ens3_present_ivan/run
/glade/derecho/scratch/asinning/ivan/new_ens4_present/marine_wrf_4.5.2/run
I am trying to run WRFv 4.5.2 on Derecho (using the pre-compiled version on Derecho), and have been running into issues during my second restart run. The model runs quickly for the initial run, and for the first restart run. However, on the second restart run, things slow down a lot, and it seems to be behaving oddly (shown in the rsl files attached below, for two different examples). The issue I see in the rsl files is when it goes to write out the first auxiliary output file (I am outputting some variables hourly, for analysis), it seemingly tries to write out the same file over and over again.
This must be why it is running slow, it is stalling on that one file. However, I cannot figure out what could be the issue here, I have toggled adaptive timestep on and off, to see if that was the issue, and the issue still persists. I also tried upping to 256 processors (originally using 128 processors) before beginning the second restart run, which did not make a difference, and the same issue persisted. There are no CFL violations, either.
I have attached namelists for two different WRF runs (different members of my ensemble, all having this issue), the corresponding rsl files, and the aux_var files. I also attached a screen capture of the rsl file during the first restart run (when the model was running well) for reference. Do you have any suggestions on why this is happening/see any issues? I have used this setup successfully for other simulations. Thank you in advance!
Here is the path to my 'ens3' and 'ens4' simulations on Derecho, if it helps:
/glade/derecho/scratch/asinning/ivan/ens3_ivan_present_wdm7rerun/ens3_present_ivan/run
/glade/derecho/scratch/asinning/ivan/new_ens4_present/marine_wrf_4.5.2/run
Attachments
-
aux_var_file_d01.txt93 bytes · Views: 0
-
aux_var_file_d02.txt417 bytes · Views: 0
-
aux_var_file_d03.txt417 bytes · Views: 0
-
namelist.input_ens4.txt7.1 KB · Views: 3
-
rsl.out.0000_ens4.txt27.2 KB · Views: 0
-
namelist.input_ens3.txt5.4 KB · Views: 1
-
rsl.out.0000_ens3.txt328.5 KB · Views: 3
-
restart_run_1_model_running_well.png323.2 KB · Views: 6