ROMS only runs with NDEFHIS set to zero

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
samphibbs
Posts: 2
Joined: Thu Oct 11, 2012 7:22 pm
Location: Imperial College London

ROMS only runs with NDEFHIS set to zero

#1 Unread post by samphibbs »

Hi,

I'm having a problem that ROMS only runs when NDEFHIS=0 in the ocean.in file. This means it produces one very large history file. I would like to produce multiple history files so my model does not encounter problems when it runs for long periods of time. But if I change the value of NDEFHIS=720 my model crashes as soon as it tries to write to the history file with the following error:

NETCDF_PUT_FVAR_1D - error while writing variable: ocean_time
in input file: /scratch/sap12/New_COAWST/Output_CFSR_2/Med12_his.nc
call from: wrt_his.F

ROMS/TOMS - Output error ............ exit_flag: 3

And no history file is produced at all. I have attached my ocean.in file. Any help on this would be really appreciated.
Many thanks,
Sam
Attachments
ocean_Med12_CFSR_SODA_Paul.in
(85.5 KiB) Downloaded 368 times

User avatar
arango
Site Admin
Posts: 1367
Joined: Wed Feb 26, 2003 4:41 pm
Location: DMCS, Rutgers University
Contact:

Re: ROMS only runs with NDEFHIS set to zero

#2 Unread post by arango »

Maybe you don't have privileges to write to the /scratch/sap12/New_COAWST/Output_CFSR_2 directory or that path doesn't exits... It is kind of suspicious that you cannot even create the NetCDF file. Perhaps, there are some problems with NetCDF library that you compiled.

samphibbs
Posts: 2
Joined: Thu Oct 11, 2012 7:22 pm
Location: Imperial College London

Re: ROMS only runs with NDEFHIS set to zero

#3 Unread post by samphibbs »

Hi thanks for the reply,

I do have permission to the directory though - both the restart and station file are created in the directory for NDEFHIS=720. Furthermore, if I set NDEFHIS=0 my model will run for full time period creating history, restart, station and average files in the directory.

The only other thing I can think of that might be both mentioning is that I have made changes to the varinfo.dat file so my forcing files are correctly read in - for most of my forcing variables I have changed the time label to 'time' eg 'pair_time' was changed to 'time' and 'tair_time' was changed to 'time' etc.

But, I'm not sure why changing the varinfo.dat file would make a difference. In fact I find it difficult to think of a problem that would only effect my run if NDEFHIS is non-zero.

Sam

User avatar
kate
Posts: 4091
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: ROMS only runs with NDEFHIS set to zero

#4 Unread post by kate »

What's odd is that it's trying to write to /scratch/sap12/New_COAWST/Output_CFSR_2/Med12_his.nc
as opposed to Med12_his_0001.nc. I would have to watch this in the debugger, or add print statements to output.F (or more).

colberg
Posts: 13
Joined: Mon Feb 16, 2004 4:54 pm
Location: CSIRO

Re: ROMS only runs with NDEFHIS set to zero

#5 Unread post by colberg »

Hi all,

has there been a solution to this problem ? I am currently having a similar issue with roms-3.6. I can use multiple output files with NDEFHIS=360 (for my particular setting) but not any other number (!), it then gives me the same error as indicated in the thread above. ROMS is trying to write to ocean_his.nc instead of ocean_his_0001.nc or so. I have not had this problem before but have been running 2d simulations mostly...but this seems rather unrelated. I wonder if this is related to changes in the roms.in file ?

ideas anyone
Frank

wzhu
Posts: 39
Joined: Sun Oct 28, 2012 6:56 pm
Location: Horn Point Laboratory UMCES

Re: ROMS only runs with NDEFHIS set to zero

#6 Unread post by wzhu »

samphibbs wrote:Hi,

I'm having a problem that ROMS only runs when NDEFHIS=0 in the ocean.in file. This means it produces one very large history file. I would like to produce multiple history files so my model does not encounter problems when it runs for long periods of time. But if I change the value of NDEFHIS=720 my model crashes as soon as it tries to write to the history file with the following error:

NETCDF_PUT_FVAR_1D - error while writing variable: ocean_time
in input file: /scratch/sap12/New_COAWST/Output_CFSR_2/Med12_his.nc
call from: wrt_his.F

ROMS/TOMS - Output error ............ exit_flag: 3

And no history file is produced at all. I have attached my ocean.in file. Any help on this would be really appreciated.
Many thanks,
Sam
Correct! I met the same problem with ROMS 3.0. In the beginning, when I set the ocean_time to zero, it worked with the non-zero NDEFHIS, but when I set the ocean_time to non-zero to change the starting time to later date, it appears the problem. but if set the NDEFHIS to zero, it works. if the history is a single file, it will be 200G in size! Do I have to set all the time parameters to the new starting time? I think it is a bug.

User avatar
kate
Posts: 4091
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: ROMS only runs with NDEFHIS set to zero

#7 Unread post by kate »

Are you really running ROMS 3.0? This has been fixed in newer versions.

wzhu
Posts: 39
Joined: Sun Oct 28, 2012 6:56 pm
Location: Horn Point Laboratory UMCES

Re: ROMS only runs with NDEFHIS set to zero

#8 Unread post by wzhu »

kate wrote:Are you really running ROMS 3.0? This has been fixed in newer versions.
Yes, but I just changed the DSTART according to the ocean_time, it works. Thanks.

Post Reply