READ_PHYPAR - variable info not yet loaded, idTTav

[linsheng]

Hi, I am running a case. I have the grid, initial, and boundary nc file. I can run this case well using the roms3.4. When i use the same include and .in filein the roms3.7, an error appear. Should I need to make some change? Error like this :
-----------------------------------------------------------------------------
READ_PHYPAR - variable info not yet loaded, idTTav

Elapsed CPU time (seconds):


ROMS/TOMS - Output NetCDF summary for Grid 01:

ROMS/TOMS - Configuration error ..... exit_flag: 5


ERROR: Illegal model configuration.

[wilkin]

You probably need to update varinfo.dat to the new version. It contains the metadata for defining the variables.

[linsheng]

Thanks for you solution, the error is eliminated! But there is another problem of NETCDF_OPEN :
Process Information:

Node # 0 (pid= 8272) is active.
Node # 4 (pid= 8276) is active.
Node # 1 (pid= 8273) is active.
Node # 2 (pid= 8274) is active.
Node # 3 (pid= 8275) is active.
Node # 5 (pid= 8277) is active.

NETCDF_OPEN - unable to open existing NetCDF file:

call from: inp_par.F, find_file

INITIAL: Configuring and initializing forward nonlinear model ...
*******

Elapsed CPU time (seconds):

Node # 0 CPU: 0.042

ROMS/TOMS - Output NetCDF summary for Grid 01:

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

Node # 1 CPU: 0.042
Node # 2 CPU: 0.044
Node # 3 CPU: 0.037
Node # 4 CPU: 0.043
Node # 5 CPU: 0.038

ERROR: Abnormal termination: NetCDF OUTPUT.
REASON: Invalid argument

[kate]

Since you had an out-of-date varinfo.dat, did you update your ocean.in as well?

This is likely to be something simple, but for me the easiest way to find it is in the debugger. Then you can step through and see which step it is actually failing on. I bet you could find this in serial mode with gdb, lacking a spendy debugger. Otherwise you can try print statements in inp_par.F and the routines calling find_file.

[wilkin]

Sometimes you get a more helpful error message if you run in serial, which you can afford to do here because the model hasn't even got to time step 1.

Sometimes in serial you will get an actual filename at ...

NETCDF_OPEN - unable to open existing NetCDF file:

... when in parallel you might not.

From the snippet of the stdout you posted it looks like ROMS is at the point where it is reading the grid, just before reading initial conditions. Check that you have specified the grid file name with the correct syntax. Do not use any TAB characters in the ocean.in file.