Cerberus

Cerberus is joining system that allows merging different kinds of records. We use it to join LIMS metadata from Pinery with workflow metadata from Víðarr.

To join data from one or more Vidarr servers with one or more Pinery servers, create a configuration file ending in .cerberus as follows:

 {
   "pinery": {
     "pinery-miso": {
       "url": "http://pinery.example.com/",
       "versions": [
         2,
         7,
         8
       ]
     }
   },
   "vidarr": {
     "prod": "http://vidarr-prod.example.com:8000"
   },
   "ignore": [
     "example-bad-provider"
   ]
 }

The "pinery" section describes all Pinery instances that can be used LIMS data sources. The keys are the provider name used in Vidarr. For each Pinery instances, multiple versions of the same data can be used by specifying them in the "versions" list.

The "vidarr" section describes all the Vidarr instances that should be used as file sources. The keys are the internal name of that Vidarr instance and the value is the URL of that instance.

The "ignore" section contains all the LIMS provider names which are present in the Vidarr instances' external keys but should NOT be merged when building file provenance. If a Vidarr workflow run contains a single external key with data from one of these ignore providers, the entire workflow run will be excluded.

To build Cerberus locally:

mvn clean install dependency:copy-dependencies

The Cerberus file provenance client can be used to produce a joined file provenance TSV in the traditional format using:

java --module-path "$(find ./*/target/ ./*/target/dependency/ \
    -maxdepth 1 -mindepth 1 -iname "*.jar" | tr '\n' :)" \
    -m ca.on.oicr.gsi.cerberus.cli/ca.on.oicr.gsi.cerberus.cli.Main online  \
    -c config.json -o output.tsv.gz