When Centos came out with php53* packages, I promptly upgraded to test them out. I did not get around to installing PECL and memcache until recently, and soon realized they were no longer available. This article shows how I was able to install PECL and memcache on a Centos (5.6) system using the IUS repository.

Since I am running the php53* packages, the provided php-pecl-memcache package is not compatible. Checking what PECL packages are available:

I decided to try out the IUS repository which provides a new set of php53 packages, along with pecl and memcache.

The IUS packages are also not compatible with the installed php53* packages, so I removed them and installed the new php53u* packages.

Next, I installed the memcached service.

After installation, I restarted Apache and checked to ensure memcache was not working.

I created a tiny script to test for memcache support:

Now, my system is ready to begin work with the Memcache API and Integration Drupal module :)

In this article, I'll show the commands I have been using to set up a fresh Centos server, configured for Apache, MySQL, PHP, Tomcat, Drupal, and Apache Solr. For my article, I used Parallels to create a virtual machine from the Centos 5.6 64bit ISOs I downloaded. To simply this article, all commands are being executed as root, firewall configurations and performance tweaks are not accounted for.

Once the distribution is installed, the first thing I do is upgrade all packages.

Install PHP, Apache, and MySQL

Set runlevels for Apache and MySQL

Install subversion. I chose to use subversion for this article because the Drupal 6.x installation works well with svn:externals to fetch the SolrPhpClient library. All subversion commands are connecting to a local subversion repository. If you are using an external server (like Beanstalk), you will have to transpose all commands from using "file://" to "https://".

Add a new local subversion repository (OPTIONAL).

Download/setup drush

Create a vhost location on the server for the Drupal installation.

Integrate the Drupal files with subversion

Download the Drupal apachesolr module

Setup MySQL

Setup Apache vhost

Reset Apache file permissions. NOTE: you will need a more solid/secure configuration for this!

Start Apache

Install Drupal via drush

At this point, you should be able to browse to your site and it will be up and running.

Now, we move onto Tomcat and Solr!

Installing Tomcat and Java. The default Centos yum repositories provide Tomcat5. I prefer Tomcat6, so there are some extras steps below and a dependency issue I had to resolve.

At this point, you should be able to access Tomcat in your browser (http://example.com:8080)

Downloading Solr Java library.

Copying the java war file into the Tomcat webapps folder will create this directory automatically:

Create Catalina config file to link war file to solr directory:

Setup Tomcat admin user(s):

Update WEB-INF/web.xml file:

Copy conf files from Drupal apachesolr module into Tomcat Solr conf directory (overwrite):

Reset Tomcat permissions/ownership:

Restart Tomcat

At this point, you should be able to access the solr/admin tomcat Page (http://example.com:8080/solr/admin)

If things are not working well at this point, check the Tomcat logs:

And, ensure the solr java module is listed in the Tomcat Web Application Manager: http://example.com:8080/manager/html

If all is well, you can now enable the Drupal apachesolr modules:

Log into your Drupal site. NOTE: default account (via drush): admin/admin

Edit default Apache Solr Host Settings:
URL: http://example.com/admin/config/search/apachesolr/server/solr/edit
Change url to: http://example.com:8080/solr, and save form.

Go to Drupal search settings page:
URL: http://example.com/admin/config/search/settings
Change the default search mode to "Apache Solr search", and save form.

Now, you are ready to test the indexing and integration.

Add a new piece of content to test indexing.
# Example:
# http://example.com/node/add/article
# title: Test Article
# Body: test test test

Browse to solr index page:
URL: http://example.com/admin/config/search/apachesolr/index
Select: Index queued content, and click on Begin button
You should see a status message like: "1 item processed successfully." and "Number of documents in index: 0 (1 sent but not yet processed)"
A few minutes later, refreshing the index page should show: "Number of documents in index: 1"

Search for "test" to verify Solr results.
URL: http://example.com/search/site/test

You can also review search results via solr/admin
URL: http://example.com:8080/solr/admin/
Enter "test" in query string box and click search

Part 2, Multicore Configuration (OPTIONAL)

If you need to run multiple sites off a single Solr Tomcat installation, you can setup multicore..

Copy the multicore xml file into your solr directory:

Create a new directory for each multisite in the solr directory:

Replicate the solr conf directory into the new multisite directory:

Update the solr.xml file:

Restart Tomcat

Now, your new multicore site will be accessible here: http://example.com:8080/solr/example.com/admin/

In this tutorial I'll show how you can setup a server to parse email with a PHP script. This tutorial assumes that your server is configured to receive email (I wrote this using a virtual machine running postfix).

The first thing you'll need to do is configure an alias to direct email to a PHP script (instead of an email box). I added the following entry to the bottom of my /etc/aliases file and then ran the "newaliases" command to refresh my aliases database:

The above entry will pipe email sent to phpscript@MYDOMAIN to the designated PHP script.

And here's the script:

At this point in the code, the body of the message will be contained in the $body variable and the headers will be in $headers.

Here is an example of the parsed headers (using print_r()):

Now, you have all the email headers and message body parsed. You can do whatever your heart desires with the data, like insert it into a database or even create nodes!

I recently got a new MacBook Pro laptop (awesome!) and went through my usual rigmarole of setting up a new [VirtualBox] virtual machine for LAMP development. In previous situations I used bridged network connections which allow my virtual machine to have its own network connection, and acquire an IP address via DHCP. Now that I plan on being more mobile, I was concerned about having a static IP address for my virtual machine (for samba connections, scripting, and any other processes that relies on a static IP address). I decided to explore alternative virtual network connections and changed my virtual NIC to use NAT (the default network connection for VirtualBox). This configuration establishes a virtual NAT for your virtual machines which is great, but the downside is I know have to setup port forwarding to connect to my virtual machine. In bridged network configurations, I could simply SSH or use Samba to connect to my virtual machine by IP address. With NAT I have to setup port forwarding for the services I need to connect to. Initially I decided to setup port forwarding for SSH (port 22) and HTTP (port 80). While my virtual machine was powered down, I executed the following commands. NOTE: you'll have to replace "Centos" with the name of your virtual machine.

Now, I can SSH to my virtual machine..

Or, connect to Apache by browsing to http://localhost:8080

The benefit of this network configuration allows me to travel anywhere with my laptop, use any type of network connection, and not have to worry about changing the way I connect to my virtual machine for development.

This morning, I encountered a PHP fatal error on my development environment. Upon further inspection, one of my third party modules (XML Sitemap) required a later version of PHP. A fresh installation of Centos 5.3 comes with version 5.1.6 of PHP. Here is an easy way to upgrade PHP to a later version by using the Utter Ramblings Yum repository.

I created a new yum repo file:

Ran a yum update:

After updating all these packages, I checked out my new PHP version:

Now, my PHP fatal error has been resolved.

NOTE: This blog entry is a re-post of a previous article.