ops-platform-setup

This repository is designed to collect all information needed to set up the Open Phacts platform. The aim is to be able to pull directly pull from github and in one go setup a running version of Open Phacts. Hopefully, this will include not only the core cache but also the other required services Something all the lines of how Heroku leverages git. In the meantime, using github will allow us to keep our documentation reasonably up-to-date as we develop.

Core Platform TODOs

In JIRA at https://openphacts2011.atlassian.net/secure/IssueNavigator.jspa?mode=hide&requestId=10701

Code

Source code needed

• https://larkc.svn.sourceforge.net/svnroot/larkc/trunk
• https://github.com/openphacts/OpsPlatform
• https://github.com/openphacts/coreGUI
• https://github.com/openphacts/OPS_LinkedDataApi

Software

• Tomcat 6
• Apache 2
• Sesame 2.6.8

Data

• Data loading script is in scripts/seasame-load-list.txt
• Run using console.sh < sesame-load-list.txt
• Each dataset is loaded into a separate named graph

Hierarchies

For each hierarchy we have three named graphs

1. the direct hierarchy - the named graph is denoted by /direct
2. the full closure - the named graph is denoted by /inference
3. other info (e.g. rdfs:label) - the named graph without a suffix

Platform set-up on ops2 (CentOS 6.2), 10 Aug 2012

Load data into Virtuoso

Set up LDA

[antonis@ops2 ~]$ sudo yum install php
[antonis@ops2 ~]$ sudo yum install php-xml
[antonis@ops2 ~]$ sudo chmod -R go+rw /var/www
[antonis@ops2 ~]$ cd /var/www/
[antonis@ops2 www]$ rmdir html
[antonis@ops2 www]$ git clone https://github.com/openphacts/OPS_LinkedDataApi.git html

Edit: /etc/httpd/conf/httpd.conf
You need .htaccess files enabled and allowed on DOCUMENTROOT

AccessFileName .htaccess

<Directory "/var/www/html">
    Options Indexes FollowSymLinks
    AllowOverride All
    Order allow,deny
    Allow from all
</Directory>

[antonis@ops2 ~]$ service httpd start

Note: The ARC library harcodes the limit on the execution time of a function call. We had to change this limit to 0 (infinite) so that we can parse large responses. The change was done in lib/arc/parsers/ARC2_RDFXMLParser.php only, because this is where the response from the triple store is parsed.

Steps to enable caching from RAM instead of disk

1. Edit deployment.settings.php from the root of LDA and change:

define('PUELIA_SERVE_FROM_CACHE', true);

to

define('PUELIA_SERVE_FROM_CACHE', false);

2. yum install php-pecl-memcache

3. yum install memcached

4. Add to php.ini:

extension=memcache.so

5. Add to/Create /etc/httpd/conf.d/memcached.conf

CacheEnable memcached / MemcachedCacheServer localhost:11211 MemcachedMaxServers 50 MemcachedMinConnections 10 MemcachedSMaxConnections 100 MemcachedMaxConnections 100 MemcachedConnectionTTL 10 MemcachedCacheMinFileSize 1 MemcachedCacheMaxFileSize 20971520 CacheDisable /admin/

6. Make sure the request Header does not contain the option 'cache-control=no-cache'