Backup

Dumping RDF data

The first thing needed for a backup is the RDF data.

Option 1: Using the Command Line Interface]

Make sure the CLI is setup correctly Note: We actually use this script to back up http://data.lod2.eu

Backing up with owcli is limited to the memory resources of the php-cli process so for huge models, direct backend backup should be used.

BACKUPDIR=/var/backups/data.lod2.eu
ONTOWIKIURL=data.lod2.eu
cd $BACKUPDIR

#iterate all models in OntoWiki
for model in `owcli -w $ONTOWIKIURL -l`
do
    #the filename here is using a hash over the graphname
    filename=`echo "$model" | md5sum | cut -d " " -f 1`
    #extract the rdf data from the ontowiki
    owcli -w $ONTOWIKIURL -m "$model" -e model:export >$filename.rdf
done

Option 2: Export directly from the store

If the store can do SPARQL and has a command line, this should always work:

Construct {?s ?p ?o} {?s ?p ?o}

Depending on the store there are more options:

Virtuoso

• http://docs.openlinksw.com/virtuoso/backup.html - Using this backup strategy will also backup the history and query cache
• http://code.google.com/p/aksw-commons/wiki/Virtuoso_ISQL
• http://www.openlinksw.com/uda/wiki/main//Main/VirtDumpLoadRdfGraphs

Backup

One way to back up the files generated above is to make a Mercurial Repository and then do hourly commits. Mercurial is file based, so you can create it locally.

Setup Mercurial

mkdir /var/backups/$BACKUPDIR
cd /var/backups/$BACKUPDIR
hg init .

Commit

hg add * 2>/dev/null
hg commit . -m "..." -q

Add a cronjob (Mentioned for completeness)

crontab mybackupscript.sh