demo.dspace.org Notes

General Server Setup / Info

Here's an overview of how everything is setup on the 'demo.dspace.org' server:

Base Software

• The server is currently running on DuraSpace's Amazon EC2 account with the following configurations:
  • Amazon AMI (EBS Image)
  • Ubuntu Linux 12.04 LTS server (64-bit)
  • Separate EBS volume mounted at /home/dspace
• All DSpace requirements/software is installed in the 'dspace' user's home directory (/home/dspace):
  • Java: /usr/lib/jvm/java-1.7.0-openjdk-amd64 (NOTE: We also have Oracle Java 7 installed at /usr/lib/jvm/java-7-oracle but it is not used by default)
  • Ant: ~/ant (installed manually)
  • Maven: ~/maven (installed manually)
  • PostgreSQL: ~/pgsql (compiled from source)
  • Tomcat: ~/tomcat(installed manually)
    • Tomcat is configured to run on port 80 (using 'jsvc' running as root)
  • DSpace Source: ~/dspace-src
  • DSpace Install: ~/dspace

'dspace' user account

• The 'dspace' user's ~/.profile initializes PATH/environment
• The 'dspace' user has limited 'sudo' access on system.
• The 'dspace' user's ~/bin/ includes various useful scripts

Server Backups

• A server snapshot (backup) is generated nightly
  
  • Snapshots from the last 4 nights are kept
  • Every Sunday a weekly snapshot is generated and kept for two weeks
  • On the first of the month a monthly snapshot is generated and kept for two months

Who to Contact

Only a DuraSpace employee can do the following:

• Install extra Ubuntu software packages on server
• Reboot server
• Upgrade Ubuntu to next version of Ubuntu
• Create Snapshots of server volumes and restore server based on one of those Snapshots

Contact sysadmin@duraspace.org or Tim Donohue if you need any of these tasks performed.

Getting SSH access to demo.dspace.org

This is how you provide a DSpace Committer with command-line access to this server.

1. Have Committer generate an SSH Key on their computer and send you their PUBLIC Key.
2. Append their PUBLIC Key on the end of the 'dspace' user's ~/.ssh/authorized_keysfile

   • NOTE: Please add a comment regarding who's key this is, so that it makes it easier to clean up later on. For example:

     # Tim Donohue's SSH Key
     ssh-rsa ....

3. They should now be able to connect as follows:

   ssh dspace@demo.dspace.org

Updating / Upgrading DSpace installation

To ensure we are consistently updating DSpace in the same manner, please perform the following steps when updating any configuration
or making any customization to DSpace.

(If you have updates/suggestions, please let us know – we can change these processes, but we just need to make sure we are all consistently following the same general steps)

Make all Configuration/File Changes in '~/dspace-src/' FIRST

The ~/dspace-src/ folder is a Git clone of the DSpace GitHub Repository: https://github.com/DSpace/DSpace/

In this local Git repository, we are running off of a local branch named "demo". You can see all the branches by running

git branch

Changes that you wish to keep should be committed to this local "demo" branch.

At any one time, you can compare this 'demo' branch to any version of DSpace. For example, to compare 'demo' to DSpace 1.8.2 run:

cd ~/dspace-src
git checkout demo
git diff dspace-1.8.2

Upgrade DSpace Source

If you are upgrading to the next stable version of DSpace, you can use git merge to help you merge all changes.

For Example:

cd ~/dspace-src
# Pull down all latest changes
git checkout master
git pull
# Merge them into our "demo" branch
git checkout demo
git merge dspace-3.0 (or 'git merge master')

NOTE: You should make sure to pay close attention to whether any Conflicts occurred. If so, you will need to resolve them manually.

Resolving Conflicts: Here are some hints on how to resolve / manage conflicts encountered during a merge:

• If there were a lot of conflicts and you just want to accept the "master" or tagged version (and overwrite any local changes), you can use:

  git checkout --theirs [full-path-to-file]
  git add [full-path-to-file]

• If you need to completely delete a file that caused conflict, just use:

  git rm [full-path-to-file]

• If you need to abort an in-process merge that had conflicts, just run:

  git merge --abort

Rebuild DSpace

cd ~/dspace-src
# Build DSpace using demo.properties and build Mirage 2 theme
mvn -U clean package -Denv=demo -Dmirage2.on=true

Push out Updates

WARNING: this overwrites existing configs in ~/dspace/config/

sudo service tomcat stop
cd ~/dspace-src/dspace/target/dspace-installer/
ant update
sudo service tomcat start

Double check everything still looks to be working.

Also make sure your changes made it to ~/dspace/ (and that you didn't remove previous settings, especially configs)

An easy way to double check config changes is to do a 'diff' of the latest dspace.cfg with the most recent '.old' one.

Recommended to commit your changes to "demo" local branch

Assuming your changes are already over in ~/dspace-src/ this is easy...

cd ~/dspace-src/
git commit [file]

# OR, to commit all changed files
git commit -a

Managing Website Content

Website / Splash page

• Tomcat is configured to run on port 80
  
  • It is using 'jsvc' to start as 'root' and then switch to being owned by 'dspace'
  • Tomcat's startup settings are all located in ~dspace/.profile
• DSpace Webapps are run from ~/dspace/webapps/ (configured in Tomcat's context fragments in ~/tomcat/conf/Catalina/localhost/)
• The main "splash" page (http://demo.dspace.org) is served by Tomcat and is located at: ~/tomcat/webapps/ROOT/index.html
  
  • Its content is also managed via the GitHub Repository at: http://github.com/DSpace/demo.dspace.org
  • Info on updating & pushing to GitHub can be found in the README at https://github.com/DSpace/demo.dspace.org

JavaDocs page

• The JavaDocs pages (http://demo.dspace.org/javadocs/) are static pages served by Tomcat and are located at: ~/tomcat/webapps/javadocs/
• These JavaDocs can be regenerated at any time by running the following (from the root source directory, [dspace-source] ):
  

  • mvn javadoc:aggregate

  • The "javadoc:aggregate" command generates a single set of javadocs which aggregate the APIs of all DSpace modules.  See http://maven.apache.org/plugins/maven-javadoc-plugin/plugin-info.html
  • NOTE: We've encountered some oddities with the results when this is run from demo.dspace.org itself (the resulting CSS isn't applied).  So, it's recommended to run this command from your local machine. It worked fine on 2 machines running Java 6, Maven 2.2.1 and 3.0.3, respectively. It didn't work on demo, which was running Java 7 and Maven 2.2.1.

Starting / Stopping DSpace (and related services)

The 'dspace' user can easily start/stop PostgreSQL and Tomcat using the corresponding service scripts:

sudo service postgresql start
sudo service tomcat start
sudo service handle-server start

sudo service tomcat stop
sudo service postgresql stop
sudo service handle-server stop

Important Cron Jobs

Obviously, you can get the latest information on the existing Cron jobs by logging into the demo.dspace.org server and running:

crontab -l

However, here's a brief overview of a few of the more important Cron jobs.

Reset DSpace Content (based on AIPs) every Saturday

EVERY SATURDAY NIGHT (currently at 23:59 UTC), all existing DSpace content is automatically REMOVED and reset to the AIPs located at ~/AIP-restore/

This is controlled by the ~/bin/reset-dspace-content script (source code in GitHub)

This is a BASH script that essentially does the following:

1. Stops Tomcat
2. Backs up current DB & Assetstore to ~/tmp/data-backup (This backup is performed just in case something goes wrong and we
   need to quickly restore DSpace.)
3. Deletes existing DB & Assetstore
4. Resets DSpace back to a 'fresh_install' state (by restoring database tables, sequences, registries, and initial Admin user acct)
5. Imports the AIPs in ~/AIP-restoreinto DSpace as default content (This also autocreates the demo EPeople and Groups)
   • SEE README in ~/AIP-restore/ for info on updating these AIPs
6. Refreshes all Indexes (Lucene/DB & Discovery) & Restarts Tomcat
7. A log of the entire 'reset' process is written to ~/AIP-restore/reset-dspace-content.log

How to update the set of demo AIPs

The set of demo AIPs are all stored in the ~/AIP-restore/ directory.

To update these AIPs, you must use the DSpace AIP Backup & Restore tools as described at:
AIP Backup and Restore

You can regenerate / update these AIPs by doing the following:

1. Install a fresh (empty) copy of DSpace on your local server.
2. Configure it to have the same handle prefix as demo.dspace.org (handle prefix: 10673) & setup an initial administrative user (ideally 'dspacedemo+admin@gmail.com' which is the Demo Administrator on demo.dspace.org).

3. Download the existing AIPs from this directory, e.g.

   scp dspace@demo.dspace.org:~/AIP-restore/* .

4. Use the downloaded AIPs to "restore" content to your local server's empty DSpace, e.g.

   \[dspace\]/bin/dspace packager -r -a -f -t AIP -e [admin-email] -i 10673/0 /full/path/to/SITE@10673-0.zip

5. Update your DSpace's content as you see fit (adding/removing/changing objects)

6. Export a fresh set of AIPs, by performing a full SITE export e.g.

   \[dspace\]/bin/dspace packager -d -a -t AIP -e [admin-email] -i 10673/0 -o includeBundles=ORIGINAL,LICENSE -o passwords=true SITE@10673-0.zip

   • The above example just exports ORIGINAL & LICENSE bundles into AIPs, and also exports user passwords into AIPs (so that they can also be restored).

7. Upload those newly updated AIPs to demo.dspace.org, e.g.

   scp . dspace@demo.dspace.org:~/AIP-restore/

   • NOTE: Before putting them on demo.dspace.org, you may want to do your own test restore using these AIPs, just to ensure there are no issues.

Reset "News" sections every night

Since the "News" sections are editable via the JSPUI, there is a cron job that automatically resets them each night.

It's a rather simple cron job that just copies the original "news-*" files from the ~/dspace-src/ directory:

05 0 * * * cp $HOME/dspace-src/dspace/config/news-* $HOME/dspace/config/ > /dev/null

Reset Demo User Passwords every hour

Since people have been known to change our demo user passwords on this demo.dspace.org server, we now reset them to the default password every hour.

This functionality is just a simple set of SQL UPDATE commands that are run via the ~/bin/reset-demo-passwords script.

kompewter IRC bot

The kompewter IRC bot is on the server at ~/kompewter.

It's source code is managed in Peter Dietz's GitHub acct at: https://github.com/peterdietz/jenni

Starting / Stopping kompewter

To start kompewter just run:

cd ~/kompewter
nohup ./jenni > kompewter.log &

(NOTE: The "nohup" command ensures that kompewter will keep running even after you log off the server.)

Java Profiling using YourKit

In order to locate potential memory issues in DSpace, we've installed YourKit on demo.dspace.org at ~/yjp/.

It can be accessed remotely so that we can perform various Java profiling tasks.

On your local computer:

• Download & Install YourKit Profiler. Put in your open source license key (available to all DSpace Committers).
• Open up YourKit, select "Connect to remote application..." option.
• Point it at "demo.dspace.org:10001" and start doing some profiling!
• If it's not running, start it using ~/yjp/bin/yjp.sh
• If needed, logs are in ~/.yjp/log/