Install ADAGUC on Red Hat 6

Required packages for the Adaguc server software

  • gcc compiler suite with g++.
  • mercurial
  • libproj-devel
  • libgd-devel
  • libcurl-devel
  • libfreetype2-devel
  • libpng-devel
  • zlib-devel
  • libxml2-devel
  • lpq-devel (postgres)
  • sqlite-devel
  • python-isodate # Optional: needed for some python scripts handling time dimensions

For Red Hat 6.2 with EPEL repositories (compiling netcdf and hdf5 not needed in this case):

yum install gcc gcc-c++ mercurial libpng-devel zlib-devel libxml2-devel \
gd-devel netcdf-devel hdf5-devel proj-devel postgresql-devel udunits2-devel \
gdal-devel cairo-devel httpd postgresql-server sqlite-devel

Required software libraries:

  • hdf5 1.8.7 or newer
  • netcdf-4.1.3 or newer

Compile instructions for hdf5

Download source code from

tar -xzvf hdf5-1.8.x.tar.gz
cd hdf5-1.8.x
./configure --prefix=<installdir>
make test
make install

Compile instructions for netcdf4

Download source code for C library from

tar -xzvf netcdf-4.x.y.tar.gz
cd netcdf-4.x.y 
./configure --prefix=<installdir> --with-hdf5=<installdir> --with-udunits --enable-netcdf-4 --enable-cxx-4 --enable-shared CPPFLAGS=-I<installdir>/include LDFLAGS=-L<installdir>/lib
make check
make install

Compile instructions for gdal

Download source code from

tar -xzvf gdal-1.8.1
cd gdal-1.8.1
./configure --prefix=<installdir> LDFLAGS="-ludunits2" 
make install

Compile the adaguc sourcecode

The source code is kept in a Mercurial repository at

Fetch the source code from the repository into a directory named adagucserver with the follwing GIT command:

git clone

The software can be built by invoking the script in the adagucserver directory. If any compile or link time information is needed this can be passed to the script through the CPPFLAGS and the LDFLAGS environment variables. For example:

export CPPFLAGS="-I/home/user/software/install/include -I/home/user/othersoftware/install/include" 
export LDFLAGS="-L/home/user/software/install/lib/ -L/home/user/othersoftware/install/lib/" 

The script builds the adagucserver executable and the h5ncdump utility in the bin subdirectory of the adagucserver source code directory. The executable has to be moved manually to the directory from where it will be invoked by the CGI script.

Create a CGI script in the http server's cgi directory

The adagucserver is invoked through a CGI script by the web server (usually an Apache web server). Find out where CGI-scripts can be stored on your system (determined by the Apache configuration) and create a CGI script for the web service

The CGI script can have the following form:

# LD_LIBRARY_PATH not needed when netcdf and hdf5 installed through yum.
export LD_LIBRARY_PATH=<installdir>/lib/:$LD_LIBRARY_PATH
export ADAGUC_CONFIG=<installdir>/config/exampleconfig.xml
export ADAGUC_LOGFILE=<installdir>/logs/exampleconfig.log
export ADAGUC_ERRORFILE=<installdir>/logs/exampleconfig.errlog

if [ "$returnCode" != "0" ]; then
  echo "Content-type: text/html" 
  echo "" 
  echo "Message:<br>" 
  echo rv: $returnCode
  if [ "$returnCode" == "143" ]; then echo "= Application was killed"; fi;
  echo "<br>" 
  echo "/Message:<br>" 

It is important to notice that this CGI script usually runs under the user id (uid) of the Apache web server. Therefore the directory where the logfiles reside (ADAGUC_LOGFILE and ADAGUC_ERRORFILE) must be writable for the web server's user id and the files if they already exist must be writable for the web server's uid.
The config file (ADAGUC_CONFIG) and it's path should also be readable for the web server's uid.