Installation
From Dpsearch
Contents |
SQL database requirements
Note that if you want to compile DataparkSearch with one of supported SQL database you must have this database already installed before installing DataparkSearch.
It is possible to use DataparkSearch with several SQL databases.
You also should have enough permission to create new database or to write into already existing one.
MySQL notes: If you want to build DataparkSearch with MySQL, 4.1 or later release required. libz library must be installed from zlib-devel RPM to successfully compile DataparkSearch with MySQL.
PostgreSQL notes: If you want to build DataparkSearch with PostgreSQL, 7.3.x or later release required. PostgreSQL 8.1 is recommended for better performance.
iODBC notes: iodbc-2.50.22a is known to work.
unixODBC notes: unixODBC-1.7 is known to work.
InterBase notes:
- Interbase 4.0 is known to work.
- InterBase CS 6.0 is known to work.
- FirebirdCS-0.9-4 is known to work.
FreeTDS notes: 0.52 version is known to work with MS SQL 7.0.
Oracle8 notes: 8.0.5.X is known to work.
Oracle8i notes: 8.1.6 R2 EE is known to work.
Supported operating systems
We use GNU Autoconf so it is possible to compile and use DataparkSearch on almost every modern UNIX system with a C compiler without any modifications. We develop the software on FreeBSD 5.x using PostgreSQL 8.1.
Currently known systems where DataparkSearch has been successfully compiled and tested on are:
- CentOS 3.1, CentOS 3.3
- Debian GNU/Linux (Lenny, etch) (i386)
- FreeBSD 2.2.5, 3.x, 4.x, 5.x, 6.x, 7.0
- Linux Fedora Core-1, Kernel 2.4.22-1.2174
- Linux Mandrake 10.2
- Linux Red Hat 8.0, 9.0
- Solaris 9
- Solaris 10 x86, gcc
- Ubuntu Linux 6.10, 7.04, 7.10, 8.04 and 8.10 i386 and amd64
- Gentoo Linux 2007.0 amd64
- OpenBSD 4.5 (i386)
We hope DataparkSearch will work on other Unix platforms as well. Please report successful platforms here.
NFS notes: There are some problems reported running DataparkSearch over NFS v4 on Linux 2.6.17. Although, everything is OK on this system when NFS v3 is used.
Tools required for installation
You need the following tools to build and install from source:
- GNU zip to uncompress the distribution.
- A reasonable tar to unpack the distribution. GNU tar is known to work.
- A working ANSI C compiler. GNU gcc is known to work.
- A good make program. GNU make is recommended and sometimes required.
- A sed - stream editor.
- A perl interpreter, if install.pl will be used for installation.
- To build documentation from XML sources, you need jade or openjade installed. You need also jadetex to be installed to build documentation in PDF. Use make book.pdf command in doc/ subdirectory to make that documentation.
Installing DataparkSearch
- Unpack the distribution and change directory into the top-level directory of the unpacked distribution. tar -zxf dpsearch-x.x.tar.bz2
- To simplify configuration process we included a configuration script with the package - install.pl. Run install.pl and select DataparkSearch configuration options in a question-and-answer manner. After you specify all the configuration options, the script will run ./configure with the options you chose. It will also create install.options file containing your configuration preferences that you can use to run the script later bypassing questions. After configuration is finished, build and install the package as described in section 3.
In case you would like to configure DataparkSearch manually without using the configuration script, do the following:
If you would like to configure the package with SQL database support:
sh$ ./configure --with-mysql
or
sh$ ./configure --with-pgsql
or with another depending on what database you prefer, or with multiple databases:
sh$ ./configure --with-mysql --with-pgsql --with-msql --with-freetds
By default, DataparkSearch is installed in /usr/local/dpsearch in the following subdirectories:
| Directory | Contents |
| bin | search.cgi, storedoc.cgi, dps-config |
| lib | libdpsearch.a(so), libdpcharset.a(so) |
| sbin | indexer, cached, run-splitter, searchd, splitter, stored |
| etc | indexer.conf-dist, search.htm-dist, langmap.conf-dist, searchd.conf-dist, stopwords.conf-dist, stored.conf-dist, storedoc.htm-dist |
| share | various documentation and sql scripts |
If you have no permission to write to that directory or just want to install DataparkSearch to another location, please use configure with --prefix option, e.g.
./configure --prefix=/user/home/data --with-mysql
To install DataparkSearch with HTTPS support use configure with the following option:
./configure --with-openssl
or in case the OpenSSL library is installed in a non-standard location:
./configure --with-openssl=/path/to/library
Note: Please note that OpenSSL library installed on your system is required for HTTPS support.
You can see all available options with ./configure --help
If you want to provide some specific flags to C compiler (for example, '-O7 -mpentium' to build highly optimized binary for Pentium™ processor if you use egcs/pgcc), you can do so using command
sh$ CFLAGS="-O7 -mpentium"
before running configure.
To compile DataparkSearch on FreeBSD with Solid in old aout format use
sh$ CFLAGS="-aout"'
before running configure.
To compile DataparkSearch on FreeBSD with aout InterBase use
sh$ CFLAGS="-aout -static"
before running configure.
You may also specify --enable-freebsd-pthreads or --enable-linux-pthreads to compile multi-threaded indexer on FreeBSD and Linux machines.
To enable DMALLOC memory debugger support use --enable-dmalloc.
The euc-kr, big5, gb2312, tscii, gujarati and shift-jis character sets are not supported by default. To built DataparkSearch with these charsets support use configure with --with-extra-charsets command line argument.
To build DataparkSearch with all additional charsets support use:
./configure --with-extra-charsets=all
To build DataparkSearch with only one specified charset support use:
./configure --with-extra-charsets=tscii
To build DataparkSearch with support for Chinese or Japanese charsets, use:
./configure --with-extra-charsets=japanese or ./configure --with-extra-charsets=chinese
To build DataparkSearch with support for several specified charsets, use a comma separated list of charsets you want:
./configure --with-extra-charset=japanese,tscii
If you run into problems with configure, please see [dpsearch-installproblem.en.html Section 2.5].
- Build and install the package.
sh$ make
sh$ make install
If you run into problems with configure, please see [dpsearch-installproblem.en.html Section 2.5].
- Create database search (for SQL database only).
You can use existing database, skip this step in this case.
MySQL:
sh$ mysqladmin create search
PostgreSQL:
sh$ createdb search
See database specific information if you use another database.
- Create sql-tables
Edit indexer.conf config file and specify DBAddr command according SQL-database used and dbmode selected. Then run:
sh$ indexer -Ecreate
indexer will create all tables automatically.
- Installing search scripts
Copy search.cgi to your web-server cgi-bin directory or make Apache alias to DataparkSearch bin directory.
Possible installation problems
- Every time you run configure, you must run make again to recompile.
To prevent old configuration information or object files from being used, run these commands before re-running configure:
sh$ rm config.cache
sh$ make clean
- If your compile fails with make errors, this can be because you are using the wrong version of make. The behavior of Solaris, FreeBSD, OpenBSD make is slightly different from GNU make. If you have make-related problems, you should use GNU make instead, often installed as gmake.
GNU make version 3.77 is known to work.
- If starting Apache with mod_dpsearch module, you're getting the following error: Undefined symbol "pthread_join" (or something similar related to pthreads), try to add the following commmand into httpd.conf file before loading mod_dpsearch.so:
LoadFile /usr/lib/libpthread.so
If above information doesn't help you, please feel free to contact DataparkSearch mailing list <dataparksearch@yahoogroups.com>
Installation registration
If you use DataparkSearch to build search on public accessible web site, you may register this site on our users page.
