Install GeneWeb in CGI mode

 Language: English • français

When daemon mode of GeneWeb is forbidden or cannot be activated, you must use CGI mode. It's mostly the case on a shared hosting account.

When running in CGI mode, GeneWeb sits behind a general purpose HTTP server such as Apache, and is launched by the server as a CGI command. As such, the first step in installing GeneWeb consists in verifying the operation of the HTTP server and of its CGI calling function.

The OS version is generally a stable version, but not the latest version. You must fit in the server environment : OS version, libraries, installed packages. Download the goode a GeneWeb package.

1&1 hosting, GeneWeb 5 and 6

The following applies to 1&1 hosting, and may differ on other hostings. On this web site, we can easily switch between two versions of GeneWeb: 5.00 and 6.00.

Directories and files

Files and directories on 1&1.
• mybases: The bases directory, beside the CGI-root directory.
• mybases.gwb
• cnt
• images
• root: The CGI-root directory.
• basesxg: An alternative bases directory.
• css: A copy of the gw/css directory. This directory is used by the Apache server.
• gw: The gw directory of the GeneWeb distribution.
• gwenv: The gw directory of the Geneweb-5 distribution.
• igw: The images directory for Geneweb-5.
• images: A directory with copies of gw/images/gwback.jpg and gw/images/gwlogo_bas.png
• pub: A directory where are readable copies of the CGI scripts (this website is a demo site).
• gw6.cgi: The CGI script which launches GenWeb-6.
• gw5.cgi: The CGI script which launches GenWeb-5.
• issue6.cgi: A test script which displays information about the environment of the server and checks size and md5sum for the gwd binary file.
• issue5.cgi: The same script, but for the GeneWeb-5 version

You need the “exec” permission on the files gw/gwd, gw/gwsetup, and files used by gwsetup (gw/gwc*, gw/gwu,gw/consang, gw/update_nldb). The gwd.arg file is empty.

The databases folder bases must be protected either by a .htaccess file, either by a location out of the HTTP server scope. It does not need to be accessible by the HTTP server.

Description of the CGI script

Main parameters:

 PWD=$(pwd) LNG="fr" GENEWEBSHARE=$PWD/gw GENEWEBDOC=$PWD/gw/doc GENEWEBDB=$PWD/../bases DAEMON=$GENEWEBSHARE/gwd LOGFILE=$GENEWEBDB/gw.log The CGI working directory. The language for the user interface. The programs folder. The documentation folder (obsolete). The databases folder. The program gwd itself. Gwd log file, it helps to solve problems.

Note that although called DAEMON, gwd does not run in -daemon mode but in -cgi mode.

Be carefull when using a log file, its size can increase quickly, don’t forget to delete it from time to time.

OPTIONS="-blang -robot_xcl 40,70 -max_clients 15 -conn_tmout 120 -min_disp_req 30 -images_url http://myserver.net/gw/images"
# -allowed_tags $GENEWEBDB/tags.txt cd$GENEWEBSHARE
./gwd -cgi  $OPTIONS -bd ../../bases > ./gwd.log 2>&1 To change your server to another version of GeneWeb (7.00 for instance), you should do the following: cd cgi-bin rm -f Gwd ln -s Gwd-7.00 Gwd  The bases folder Folder structure for GeneWeb bases (Genealogy data). The base folder should contain the same information structure as in the case of installation on a personal computer: • one or several base.gwb folders. • one or several base.gwf files. • cnt containing access count data. • etc containing template text files used in priority over the generic gw/etc files (see header). • images containing for each base photos associated with each person (first_name.occ.last_name.jpg). • lang containing some base specific language relates template text files (lexicon for instance). • src containing for each base text files and image files inserted in notes with the m=SRC command. • wiz.auth as many wizards and friends authorization files as specified for each base in their respective .gwf parameter file. (Note that the picture on the right has been taken off a personnal computer where the bases are stored in a folder names GeneWeb-Bases, as opposed to bases in the remote server structure proposed here). Script to upload a copy of a base (Test and report for errors or improvements are welcome.) The script base-upload.sh (download it) takes a base_name as an argument, and optionally images or src. It uploads a fresh base on your server, or a fresh copy of the images/base or src/base folders. It performs the following steps: • creates a fresh base.gw file from the base on your personal computer. • extracts a listing of the content of bases/images/base bases/src/base and bases/src/base/images in three ls-xxx.txt files. • creates a remote.sh file (see below). • creates a tar file containing$BASE.gw, history remote.sh and the three ls-*.txt files above.
• sends the tar file to the remote server.
• triggers on the remote server unfolding of the tar file and execution of remote.sh. The result of this remote execution should be a mail containing the remote.log and three diff between images folders.

This procedure saves the previous folder of the base or its images and src folders with a "yyyy-mm-dd-hh:mm:ss" date tag. As time progresses, you may want to clean-up this accumulation of saves.

remote.sh for base update

The shell script below is executed on your remote server, and installs your base at the proper location (according to the overall set-up described here).

#!/bin/sh
# BASE BASES_SERVER BIN_DIR_SERVER and ADDRESS will be replaced by the appropriate values by sed
cd
DATE=$(date +"%Y-%m-%d-%T") echo 'Error log of update for base: BASE' > ~/remote.log echo$DATE >> ~/remote.log

# Save current version
cd BASES_SERVER 2>> ~/remote.log
if [ -d ./BASE.gwb ]  # do it only if folder exists
then
mv ./BASE.gwb  ./BASE-$DATE.gwb 2>> ~/remote.log fi # Extract new base from .tar file tar xf ~/BASE.tar 2>> ~/remote.log # rebuild BASE from .gw file BIN_DIR_SERVER/gwc1 -f -o BASE BASE.gw 2>> ~/remote.log BIN_DIR_SERVER/updnldb BASE 2>> ~/remote.log if [ -f ./BASE-$DATE.gwb/history ] # do it only if file exists
then
cp -f ./BASE-$DATE.gwb/history ./BASE.gwb 2>> ~/remote.log fi if [ -f ./BASE-$DATE.gwb/forum ]  # do it only if file exists
then
cp -f ./BASE-$DATE.gwb/forum ./BASE.gwb 2>> ~/remote.log fi # done cd >> ~/remote.log # Create a link to css.txt in the bases/src folder ln -s -f ~/GeneWeb/geneweb-$VERS/gw/etc/css.txt BASES_SERVER/src/BASE/css.txt 2>> ~/remote.log

# compute diffs between server and personal computer for image folders
echo '' >> ~/remote.log
if [ -d BASES_SERVER/images/BASE ]     # do it only if folder exists
then
ls BASES_SERVER/images/BASE          > ./ls-personnes-serveur.txt
echo 'Personnes diff serveur vs local' >> ~/remote.log
diff ./ls-personnes-serveur.txt ./ls-personnes.txt >> ~/remote.log
else
echo "No folder BASES_SERVER/images/BASE" >> ~/remote.log
fi
echo '' >> ~/remote.log
if [ -d BASES_SERVER/src/BASE/ ]       # do it only if folder exists
then
ls BASES_SERVER/src/BASE             > ./ls-src-files-serveur.txt
echo 'Src files diff serveur vs local' >> ~/remote.log
diff ./ls-src-files-serveur.txt ./ls-src-files.txt >> ~/remote.log
else
echo "No folder BASES_SERVER/src/BASE" >> ~/remote.log
fi
echo '' >> ~/remote.log
if [ -d BASES_SERVER/src/BASE/images ] # do it only if folder exists
then
ls BASES_SERVER/src/BASE/images      > ./ls-images-serveur.txt
echo 'Images diff serveur vs local' >> ~/remote.log
diff ./ls-images-serveur.txt ./ls-images.txt >> ~/remote.log
else
echo "No folder BASES_SERVER/src/BASE/images" >> ~/remote.log
fi

rm ./ls-*.txt

/usr/bin/mail -s 'Update error log' 'ADDRESS' < ~/remote.log
rm -f ~/remote.*

remote.sh for images or src update

The shell script below is executed on your remote server, and installs the images or src/images folders for your base at the proper location (according to the overall set-up described here).

Depending on the size of your images folders, and on the number of new images requiring upload, you may prefer to upload images individually with your preferred ftp client rather than doing the bulk upload proposed here. Remember that the base upload script performs a comparison between the images folders on your personal computer and your server.

#!/bin/sh
# BASE BASES_SERVER and ADDRESS will be replaced by the appropriate values by sed

DATE=$(date +"%Y-%m-%d-%T") if [ -e ./images.tmp ] then FILES="images" rm -f ./images.tmp fi if [ -e ./src.tmp ] then FILES="src" rm -f ./src.tmp fi echo "Error log of update for$FILES of base: BASE" > ~/remote.log
echo "date" >> ~/remote.log

# Save current version then move new version
if [ -d BASES_SERVER/$FILES/BASE ] # do it only if folder exists then mv BASES_SERVER/$FILES/BASE  BASES_SERVER/$FILES/BASE-$DATE 2>> ~/remote.log
fi
mv ./BASE BASES_SERVER/$FILES 2>> ~/remote.log /usr/bin/mail -s 'Update error log' 'ADDRESS' < ~/remote.log rm -f ~/remote.* Debugging your remote server Debugging your remote server may be tricky, but some systematic approach and several tools will help. • Verify first that your HTTP server works properly. This is achieved by typing your_server_name in the URL window of your browser which should return the content of index.html. • Verify that the cgi mechanism works. This is achieved by typing your_server_name/cgi-bin/test-cgi.sh as seen above (see #Folders and files). If your server returns a "Internal server error", several hypothesis should be examined: • Your script test-cgi.sh does not work properly. Run it directly on a terminal window by typing ./test-cgi.sh (see #Folders and files). • Your server accepts only files with extension .cgi. Rename test-cgi.sh into test-cgi.cgi and try again. • The first two lines returned by your test-cgi.sh script are not exactly as shown (the second one should be a blank line, Note also that "Content-type text/html\n\n" did not work in my tests). • You also may want to verify that your "End-of-Lines" are correct for your server environment. Remember that there are three different such EOL encoding for Windows, Mac, and Linux!! • Examine the HTTP server log file on your server by typing: tail ~/logs/access.logs.current (your environment may store access logs somewhere else, but this is the most likely place) Examination of the log file may give you a hint as to the nature of your problem. • Another typical issue is that of ownership of the various files associated with GeneWeb. In the case described here, ownership is user and group ownership is ftpusers. Depending on the specific method you have used to install GeneWeb, this may vary. Some discussions on the Yahoo! group forum about GeneWeb mentions geneweb for group ownership! • You may also want to look at the gwd log file if your GeneWeb server works only partially (welcome page is ok, but other pages do not work properly). This log file is specified in the gwd launch command. In our example, it sits at $DIR/gwd.log where \$DIR depends on the specific version of GeneWeb you are currently running (see details in #The cgi folder section above).
• You may end up in a situation where GeneWeb works only partially, in particular it may miss some css style sheet or JavaScript files. One way of debugging this kind of problem consists in exploring the HTML source file produced by GeneWeb (your browser offers this capability addressed to "developers"). Is such source files, you will find reference to css and JavaScript files whose source should appear in your browser if you click on the link. In case of bad configuration, you will obtain an error messages such as "File not found" or "You do not have permission"…
Remember that some of the files needed for proper display are fetched by the HTTP server rather than by GeneWeb.

Access rights and protections

In CGI mode, gwd runs behind a standard HTTP server, typically Apache. As such, the owner of the gwd process is the owner of the HTTP server. For instance, with Apache, this owner is defined in the http.conf file and its default value is set to user: _www and group: —www. When running within a hosting environment, such as 1&1, you typically do not have access to this parameter. You must therefore organise the protection level of the various folders involved with GeneWeb appropriately:

• read access is usually allowed by default
• for wizards, gwd needs write access to bases/cnt/actlog, bases/cnt/robot (if you have activated the -robot_xcl parameter)
• and when modifications are performed, gwd needs write access to bases/basename.lck, bases/basename.gwd/notes_link and to bases/basename.gwd/patches

Creating those files, and doing a sudo chown _www filename seems to be sufficient.

Access problems are reported in the HTTP error log file, unfortunately not always available in the case of shared hosting services!!

Access to various files through your hosted HTTP server is also controlled by your hosting service and through .htaccess files spread across your folders. Managing a coherent set of .htaccess files is not trivial and error prone!! The examples below apply to Apache version 2.4 only (earlier versions have different directives!!).

Options +FollowSymLinks should allow following symbolic links

Require all denied prevents access to this folder to all users. Require all granted allows access to this folder to all users.

AllowOverride AuthConfig
AuthType Basic
AuthUserFile /Users/Name/SomeFolder/htfriends.auth
Require valid-user


should restrict access to this folder to users supplying a valid password as defined by the htfriends.auth file.

htpasswd /Users/Name/SomeFolder/htfriends.auth UserName will trigger the process to add a new user to the list.

GeneWeb Manual

Use and manage genealogical databases

Technical annex