I was also getting the error
Fixx_rofi is not a StandardName in the NUOPC_FieldDictionary!
so I tried your suggestion of changing to
modules:
use:
- /g/data/vk83/modules
load:
- access-om3/2025.01.0
- nco/5.0.5
Which seems to have got me past the Fixx_rofi error, but using this version I am now getting an error where the model is not reading in the full length of what I need it to in the mom override file so I get the error
FATAL from PE 0: There is a mismatched quote in the parameter file line: #override OBC_SEGMENT_001_DATA = "U=file:forcing_obc_segment_001.nc(u_segment_001), V=file:forcing_obc_segment_001.nc(v_segment_001),SSH=file:forcing_obc_segment_001.nc(eta_segment_001),TEMP=file:forcing_obc_segment_001.nc(temp_segment_001),SALT=file:forcing_obc_segment_001.nc(salt_segment_001),Uamp=file:tu_segment_001.nc(uamp),Uphase=file:tu_segment_001.nc(uphase),Vamp=file:tu_segment_001.nc(vamp),Vphase=file:tu_segment_001.nc(vph
I had this problem in the past with some executables but not others. I think somewhere in the MOM6 code there is a buffer value which needs to be longer?
Hi Lizzie
This error “sounds” like its just an issue in how MOM_override is formatted? e.g. a quotation mark is missing? Can you push your config to github and we can look?
Hi Anton
Here is my config MOM6-regional-BGC/MOM_override at main · elizabeth-ellison/MOM6-regional-BGC · GitHub
I can get around the FATAL from PE 0: There is a mismatched quote in the parameter file line: error by splitting the line (eg line 17 in the override file) intoshorter parts for each of the 4 segments
But I then get
Exit code: 66 forrtl: severe (66): output statement overflows record, unit -5, file Internal Formatted Write”
I had the same error before and fixed it using a different executable!
I guess the OBC_SEGMENT_001_DATA line is too long
This comment implies you can split the line with a \ or &
I haven’t tried that, are you happy to give it a go and see if it works?
Hi @Lizzie thanks for brining this one up! As this topic has come up a couple of times before I think it is worthy of a new thread. It will make it easier for the next person with the issue to find the solution.
Hi Anton
Splitting the #override OBC_SEGMENT_004_DATA = "U=file:forcing_obc_segment_004.nc(u)," // & "V=file:forcing_obc_segment_004.nc(v)," // & ...
seemed to work because I’m not getting that error message anymore! But now the run just seems to be hanging without any error messages or outputs. The status of the job is just stuck at 00:00:06 R even after almost an hour.
Thanks Lizzie. I’m reaching the limits of my knowledge here. Does it run if you specify less boundary forcings (is that possible with the rest of the configuration?) I’m hoping @angus-g or @ashjbarnes have ideas ?
I have just gotten the same issue – but without changing mom_override. So I don’t think that this error was to do with the boundary conditions. The “time use” is what is reported with qstat:
>qstat
Job id Name User Time Use S Queue
--------------------- ---------------- ---------------- -------- - -----
137902938.gadi-pbs 1deg_jra55do_ia **** 00:00:06 R normal-exec
Time Use is stuck at 00:00:06 for a long time. I would expect Gadi to still increment the clock if we had done something to cause the model to hang. Either way, it looks like the boundary conditions is fixed, and we are onto the next error - which is a weird one!
Hi Lizzie
I didn’t have the same issue using the express queue, instead of normal but I also didn’t try normal (noting express costs 3x as much as normal, so more suited to debugging and development etc).
Your config didn’t run for me due to the processor layout. The ocean_ntasks in nuopc.runconfig was set wrong. The easiest thing was just to turn off the mask table and use 96 cores (nodes on gadi are 48 cores, and you have to use a whole node, so best to use a multiple of 48 cores).
With set 96 cores · anton-seaice/MOM6-CICE6@de935d6 · GitHub
it felt like progress and forcing_obc_segment_001.nc appears to load ok, the other files give errors like this:
init_extern_field: field u)_segment_002 not found in INPUT/forcing_obc_segment_002.nc.
Hopefully that’s useful!
Hi @anton
Thanks for checking that. @helen and I think maybe it was just a Friday afternoon Gadi issue because it just ran fine for us both in the normal queue.
I had to add {} under userscripts: or payu wasn’t happy.
But then I also got this issue
"FATAL from PE 2: MOM_domains_init: The product of the two components of layout, 10 10, is not the number of PEs used, 48."
so will try your suggestion now!
Hi @Lizzie
You’ll need to request at least 100 CPUs for a 10x10 layout as MOM6 requires the product of the layout = number of processors used.
See here for more info
https://forum.access-hive.org.au/t/access-rom3-setup-instructions/4107/58?u=paul.gregory
The payu launch script will automatically bump this to 144 (i.e. 3 x 48) for reasons @anton mentioned above.