Can't run WRF from wrfrst files

[jkaufman]

I am trying to run a hi-res simulation over a span of several days. The initial run of the model, where restart = .false., runs perfectly fine and produces restart files at the desired time for each of the 3 domains I am running. However, when I set this parameter to .true., I always receive the 'forrtl: severe (66)' error. It doesn't appear that the debug lines really help in determining what is actually wrong with this, though I've gathered from glancing at other forums that this relates to some sort of overflow. Do I have an issue in my namelist?

Thank you in advance for any advice you have to offer.

 

[kwerner]

Hi,
You're right that debug_level isn't typically very useful. Can you try to run this again, setting debug_level = 0, and then package up all of the *.out and *.error files (e.g., rsl.error.*) into a single *.tar file, and attach that file, along with the updated namelist.input file? Thanks!

 

[jkaufman]

Hello,
Thank you for your reply. It turns out that the issue is related to the 'rst_inname' line, where the overflow occurs due to a path string that surpasses a certain limit. This has been mentioned in a previous forum post, which is linked below:

WRF restart error caused by "too long" directory specified by "rst_inname"

Issue: WRF restart would fail if the restart files are specified by the namelist variable "rst_inname" that contains a directory with a length greater than 17. The following error message can be thrown out by the model wrf forrtl: severe (66): output statement overflows record, unit -5, Cause...

forum.mmm.ucar.edu

Thanks again for your time!

 

[kwerner]

Thanks for letting me know. Were you able to use the solution provided in that post to help you get past the issue?

 

[yliu52]

Hello, I met a similar issue. I am running high resolution (WRF LES) using WRF 4.4 over a span of two days. The initial run of the model, where restart = .false., runs ok and generated restart files after 12 hours each of the 6 domains. But, when I set restart to .true., and changed the simulation is always stuck/frozen at the beginning as shown below, however, the computation time (wall time) is still running. I am using NAS pledias system. The namelist is attached here. Could you please help me take a look? Thanks a lot for all your help in advance.
tail rsl.out.0000
ips,ipe,jps,jpe 1 13 1 13
INTERMEDIATE domain
ids,ide,jds,jde 36 74 36 74
ims,ime,jms,jme 31 51 31 51
ips,ipe,jps,jpe 34 41 34 41
*************************************
d03 2023-05-26_03:00:00 alloc_space_field: domain 4 , 8860572 bytes allocated
d03 2023-05-26_03:00:00 alloc_space_field: domain 4 , 62070288 bytes allocated
RESTART: nest, opening wrfrst_d04_2023-05-26_03:00:00 for reading
d03 2023-05-26_03:00:00 Input data is acceptable to use: wrfrst_d04_2023-05-26_03:00:00

 

[kwerner]

@yliu52
Hi, Since you are only running a 2 day simulation, do you need to do a restart? I understand if there's a reason why you need to, but I just want to check before we dig into this. Thanks!

 

[yliu52]

Hey Kwerner,

Thank you for your reply. As the longest walltime is 120 hours and the time step is 9s for six domains, restart is needed. Also, The first 24 hours as spin-in time. I would have to use restart after 12 hours for twice, then change the history_interval to get the target time written to the output file.

Thanks a lot,
Yunsong

 

[kwerner]

You should only need 9-12 hours for spin-up, but I assume you would still need to restart, even if doing that?

If so, can you share all of your error/output files? If you have several rsl.* files, please package them all together into a single *.tar file and attach that.

A few tests you can also try:
1) Check your disk space to make sure you have enough to write more files.
2) Try to see if any one of the domains is causing the issue - you can test a restart run with only a single domain, then 2 domains, then 3 - as long as it works, until you get to the one that fails.

After that, in addition to the rsl files, can you also share the restart file for each domain, along with the wrfbdy_d01 file so that I can test it here? Those files will be much too large to attach, so take a look at the home page of this forum for instructions on sharing large files.

 

[yliu52]

Hey Kwerner,
Thank you so much for your replies. I was changing emails with NAS support team, they suggested me to redo the simulation to generate new restart files. I have tried their suggestion for one simulation. I turned on the restart function today, it is running. I am moving on for other simulations. It looks like the problem is solved after using the re-generated restart files.

Thanks a lot,
Yunsong

 

[kwerner]

That's great news! Thanks for the update.