[ncl-install] ncattname error when reading NetCDF files

[Mary Haley]

Hi  Samuel,

>From your compilation list below, it looks like you did everything
correctly. It's possible there's an issue with NetCDF V4.0, but we
haven't heard of anything like what you describe below.

I see Dennis has asked you for the file offline, so we'll take a look
at it.

--Mary

On Wed, 8 Oct 2008, Samuel G Trahan wrote:

>
> I am trying to get NCL working on the IU BigRed cluster.  I was able to open 
> and view a few grib files, and create X11 graphical output. However, when I 
> went to open a NetCDF file, I got a few hundred lines of errors that look 
> like:
>
> 	ncattname: ncid 65536: NetCDF: Not a valid ID
> 	ncattinq: ncid 65536; varid 10; attname "ST100200": NetCDF: Not a 
> valid ID
>
> and one "fatal:NclMalloc Failed:[errno=12]" error. Then, when I use 
> print(file), it lists out all of the variables in the file and says they have 
> size 0.
>
> The script I'm using contains:
>
> begin
>  file1=addfile("/N/u/tg-strahan/BigRed/work/katrina-gfs/run/met_nmm.d01.2005-08-27_00:00:00.nc","r")
>  print(file1)
> end
>
> The file is output from the WRF Preprocessing System (WPS) version 3.0.1. I 
> also tried some wrfout_* files output by the atmospheric simulator WRF-NMM 
> 3.0.1.1.  According to ncdump -h, the variables do not have size zero (and 
> the files are tens of MB to 1.3 GB in size).
>
> The error I'm getting is not the ncclose error mentioned in the archive.  The 
> error is happening when reading the file, not closing it.
>
> Do you have any idea what might be causing this problem?  I've included 
> details below if you need more info.
>
> Thank you in advance,
> Samuel G Trahan
>
> The output generated by the addfile line is as follows:
>
> -------------------- begin addfile errors ---------------------
> ncattname: ncid 65536: NetCDF: Not a valid ID
> ncattinq: ncid 65536; varid 10; attname "ST100200": NetCDF: Not a valid ID
> ncattname: ncid 65536: NetCDF: Not a valid ID
> ncattinq: ncid 65536; varid 10; attname "ST100200": NetCDF: Not a valid ID
> ...
>    more lines like that
> ...
> ncattname: ncid 65536: NetCDF: Not a valid ID
> ncattinq: ncid 65536; varid 12; attname "ST010040": NetCDF: Not a valid ID
> Copyright (C) 1995-2007 - All Rights Reserved
> University Corporation for Atmospheric Research
> NCAR Command Language Version 5.0.0
> The use of this software is governed by a License Agreement.
> See http://www.ncl.ucar.edu/ for more details.
> fatal:NclMalloc Failed:[errno=12]
> ncattget: ncid 65536: NetCDF: Not a valid ID
> ncattname: ncid 65536: NetCDF: Not a valid ID
> ncattinq: ncid 65536; varid 12; attname "ST010040": NetCDF: Not a valid ID
> ncattname: ncid 65536: NetCDF: Not a valid ID
> ncattinq: ncid 65536; varid 12; attname "ST010040": NetCDF: Not a valid ID
> ... more lines like that ...
> --------------------- end addfile errors ----------------------
>
> (Except, without the "begin addfile errors" and "end addfile errors" lines.) 
> Those errors are printed out even if I remove the print(file1) line from my 
> script.  Running print(file) results in the following lines being printed:
>
> ------------------- begin print(file1) output -----------------
> Variable: file1 (file variable)
>
> filename:       met_nmm.d01.2005-08-27_00:00:00
> path: 
> /N/u/tg-strahan/BigRed/work/katrina-gfs/run/met_nmm.d01.2005-08-27_00:00:00.nc
>   file global attributes:
>      XLAT_M : <ARRAY of 0 elements>
>      XLAT_M : <ARRAY of 0 elements>
>      XLAT_M : <ARRAY of 0 elements>
> ... a bunch of those XLAT_M lines ...
>   dimensions:
>      Time = 1  // unlimited
>      DateStrLen = 19
>      west_east = 179
>      south_north = 323
>      num_metgrid_levels = 27
>      num_sm_levels = 4
>      num_st_levels = 4
>      z-dimension0012 = 12
>      z-dimension0016 = 16
>      z-dimension0024 = 24
>   variables:
>      character Times ( Time, DateStrLen )
>
>      float PRES ( Time, num_metgrid_levels, south_north, west_east )
>         PRES : <ARRAY of 0 elements>
>         PRES : <ARRAY of 0 elements>
>         PRES : <ARRAY of 0 elements>
>         PRES : <ARRAY of 0 elements>
>         PRES : <ARRAY of 0 elements>
>
> ... the rest of the variables printed with those "0 elements" messages ...
> ------------------- end print(file1) output -------------------
>
> According to ncdump -h, these files are valid -- the variables do not have 
> zero size.  I am using the IU BigRed Linux PowerPC cluster with suse 9, patch 
> level 4 and the IBM XL compiler.  (I had to use gcc for some of the 
> prerequisite libraries though.)  The versions of everything I compiled were 
> as follows:
>
> ncl_ncarg_src-5.0.0 (plus triangle.h and triangle.c)
> HDF4.2r3 (configured with --disable-netcdf to avoid undefined sd* symbols)
> hdf5-1.8.1
> g2clib-1.1.7
> g2lib-1.1.7
> szip-2.1
> netcdf-4.0 (configured with --enable-netcdf-4 and --with-hdf5)
> openssl-0.9.8i
> curl-7.19.0
> libxml2-2.7.1
> libdap-3.8.2
> libnc-dap-3.7.3
> vis5d+-1.2.1   (vis5d was linked with netcdf-3.6.3 instead of 4)
> udunits-1.12.9
> openldap-stable-20080813 (required by libcurl)
>
> _______________________________________________
> ncl-install mailing list
> ncl-install at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-install
>