Time/Place
This meeting is a hybrid teleconference and IRC chat. Anyone is welcome to join...here's the info:
- Time: 11:00am Eastern Daylight Time US (UTC-4)
- Google-Hangout
- IRC:
- Join the #duraspace-ff chat room via Freenode Web IRC (enter a unique nick)
- Or point your IRC client to #duraspace-ff on irc.freenode.net
Attendees
Agenda
- Multi-document Indexing
- Fedora ontology
- Clustering status
- other...
Minutes
Multi-document indexing
- Early on, we realized rich indexing is important
- Indexing should be externalized
- Eric James raised the issue that currently there is just a one-to-one mapping of index-document to resource
- Goals
- Persist multi-mapping in the repository
- Maintain architectural boundaries
Fedora ontology
- We need to more clearly define Fedora's three ontologies
- We also need to make the elements of the ontologies better defined
- To what extent are our ontologies independent from the JCR notions
- Art Inst. of Chicago will be publishing their ontology soon
- Question on the need (or lack thereof) built-in dependencies to Dublin Core
- Possibly use transform module to present properties
- Notions: persisting, presenting, inferring assertions
General Search
- What will be the most reliable search mechanism
- Chicago is begining to build a client
- Recommendation is to go external
- All three are in the same tomcat (fcrepo, jms-indexer, sesame)
- Andrew to create ticket: restarting fedora makes tomcat hang
Clustering
- Was able to get a configuration working with "distributed" node
- not exactly performant, but no errors
- working on local network, will test on SCC next
- Results at the moment
- 200mb/s (20 threads) - single
- 2mb/s (20 threads) - cluster
- Will be creating a github project with the configuration
- Two weeks, need to present fcrepo (FIZ)
- Request to have others test Frank's posted configuration
- Stefano may be able to test this
Stefano
- Interest in dissemination
- Automated generation of derivatives on ingest
- Working on an external data-transformation service
- metadata, thumbnails, etc
- Looking at an event model
- Sequencers seem like a good approach
- They are not exactly flexible
- Interest in JMS and/or webhooks
- There is also a separate goal of interest
- Create a public repo that mirrors only part of the internal repo
- Could permissions help here? or separate workspaces? or a mirrored repo?