Running DSpace

A technical talk about CGSpace in the CGIAR, and what it takes to keep it up and running (including the people, skills, tools, and workflows).

1. Running DSpace Technical overview, lessons learned, workflows and essential skills October, 2013
2. DSpace Instances Hosted at CGNET in California, USA CGSpace DSpace Test One server, two instances...
3. Instance Overview CGSpace (cgspace.cgiar.org) • “Production” • Should always be up & stable • Is the “reference” implementation DSpace Test (dspacetest.cgiar.org) • “Development” • Changes to style, functionality, DSpace etc are tested here first • Sometimes wiped clean
4. Two DSpaces, one server!
5. Living With Legacy Decisions... CGSpace and DSpace Test on the same machine… • In 2010 CGSpace had a fraction of the content, users, etc, so it didn’t affect the running of the system • Not true anymore! • 100s of 1000s of monthly views... • Large assetstore, log files, RAM / CPU usage, etc
6. (Near) Future Plans cgspace.cgiar.org dspacetest.cgiar.org Separate instances on Amazon EC2!
7. CGSpace Code Is 100% Open Source code is on github: github.com/ilri/DSpace
8. How The Code Is Organized Production code lives in the 3_x-prod branch; this is stable, tested code. Updates (if any) come from the development branch on Monday. Development code lives in the 3_x-dev branch; this is semi-tested code! Changes throughout the week.
9. “Social Coding” on GitHub • Anyone can “fork” the code repository to their own GitHub account • Source code repositories can share code via “pull requests” • Developers can comment on changes and discuss issues GitHub “OctoCat”
10. “Pull request” from @mire introducing the CUA module
11. All changes were not created equal...
12. • Sending changes is good, but leaves the burden of merging to me • Sending patches is better, but requires sender to know how to generate them • Sending a pull request is best, but requires sender knows how to use git, branches, etc Workflow Lessons Learned
13. Go Forth And Fork! … and send pull requests!
14. Scenario: Create A New Theme Creating an XMLUI theme for a new community 1. Create community in DSpace (ie, 10568/38440) 2. Add custom metadata (ie, cg.subject.bioversity) 3. Add custom submission template (input-forms.xml) 4. Copy existing XMLUI theme (ie ILRI) as a reference, and customize for center-specific metadata, look & feel, etc 5. Update search & browse indexes (dspace.cfg) 6. Update XMLUI config for new theme (xmlui.xconf)
15. DSpace Sysadmin Crashcourse DSpace... • is a Java application • builds using maven and ant • uses PostgreSQL as a database backend • stores PDFs and other blobs in the filesystem (“assetstore”) • runs best on Linux
16. CGSpace Stack Apache httpd Apache Tomcat PostgreSQL Bitstreams Debian GNU/Linux
17. Why Not Use Tomcat Directly? Any sysadmin will tell you that working with Tomcat is a joy*. Surprisingly**, these things are annoying in Tomcat: • Virtual hosting • SSL • redirects • caching and manipulating headers *for some definitions of “joy” **not surprising, actually
18. Essential Technical Skills Managing a DSpace instance doesn’t require “programmers” or “developers” (but it doesn’t hurt). Mainly, you’ll need: • Linux experience (Debian, CentOS, Ubuntu) • Administration experience (web servers, log files, cron jobs, security) • Software development concepts (git, patches, branching/merging)