Coreseek-3.2.13 install in CentOS-6.2

It is never easy to install Coreseek fulltext-search source-codes. The Coreseek’s installation documents are not very clear, and its installation largely depends on environment settings. So it’s common having problems to install Coreseek.

After smoothly installing Sphinx 2.0.5, SCWS-1.2.0, I finally installed successfully Coreseek-3.2.13 in CentOS 6.2. Here I summarized the steps.

1. pre-install support packages.

Read the pre-install requirements carefully and check what missed on the server. I use ‘yum’ to install whatever I needed.

$ sudo yum install mysql-devel libxml2-devel expat-devel imkae gcc-c++

You can use ‘sudo yum install’ to install other packages such as ‘Development tools’, ‘php-devel’ etc if needed.

2. Dowload coreseek 3.2.13

Downlad Coreseek, which include 3-parts, and need to install 1 by 1:

$ wget http://www.coreseek.cn/uploads/csft/3.2/coreseek-3.2.13.tar.gz $ tar xzvf coreseek-3.2.13.tar.gz $ cd coreseek-3.2.13

3.Install mmseg part

$ cd mmseg-3.2.13

If you run ./bootstrap directly, the first command: aclocal -I config will fail: config/sys_siglist.m4:20: warning: underquoted definition of SIC_VAR_SYS_SIGLIST

Here is 2 small tricks to fix it: (a) edit bootstrap file, change shabang from #!/bin/sh to #!/bin/bash to inherit path and env variables. (b) use root instead to execute. $ sudo ./bootstrap

It should pass, then use ‘root’ permission to do configure & make: $ sudo ./configure —prefix=/usr/local/mmseg $ sudo make $ sudo make install Without ‘sudo’, there probably throw a lot of warnings and errors.

4. Install Coreseek part

$ cd csft-3.2.13/

Again, I change buildconf.sh’s shabang from #!/bin/sh to #!/bin/bash.

$ sudo bash buildconf.sh

$ ./configure —prefix=/usr/local/coreseek —without-unixodbc —with-mmseg —with-mmseg-includes=/usr/local/mmseg/include/mmseg/ —with-mmseg-libs=/usr/local/mmseg/lib/ —with-mysql $ make; sudo make install

This way it compiled with mysql, and mmseg library.

5. Test by testpack

$ cd testpack/ $ locale to check locale’s LANG’s setting, do the following setup: $ export LANG=zh_CN.utf-8 $ export LC_ALL=zh_CN.utf-8

You may put these variables into /etc/profile.d/lang.sh or $HOME/.bash_profile or /etc/sysconfig/i18n or $HOME/.i18n.

$ /usr/local/mmseg/bin/mmseg -d /usr/local/mmseg/etc var/test/test.xml The correct word segments display. That is cool.

From it’s README (http://www.coreseek.cn/products-install/install\_on\_bsd\_linux/), it needs higher version of m4,automake,autoconf (in CentOS 5.5) which I don’t have. My platform is CentOS 6.2(not CentOS 5.5), and the versions of these tools are lower than their requirements. However, it still works.