-
Notifications
You must be signed in to change notification settings - Fork 71
January 27, 2021
This meeting is held virtually via Zoom, with an open channel for chatter on Slack. Anyone is welcome to join. Here is the info:
- Time: 1:00pm Eastern Standard Time US (UTC-5)
- Zoom link: https://zoom.us/j/97345581392
- Join us on the Slack Islandora 8 Channel https://islandora.slack.com/archives/CM5PPAV28
- Danny Lamb
- Melissa Anez
- Don Richards
- Kristina Spurgin
- Alexander O'Neill
- Jordan Dukart
- Brandon Weigel
- Alan Stanley
- Willow Gillingham
- Seth Shaw
- Nat Kanthan
- Adam Vessey
- Kyle Huynh
- Bethany Seeger ⭐ 💪
- Open Meeting Tomorrow
- Fedora 6 Updates
- Not using Etags, but "State Tokens"
- https://jira.lyrasis.org/browse/FCREPO-3613
- Triplestore Indexer
- Issue Roundup:
- PR Roundup:
- Advance Queue + Triplestore indexing (https://www.drupal.org/project/triplestore_indexer) demo - Kyle
- Feel free to add agenda items...
(Last 10 Minutes): Roundtable
11 Atlantic time - Open meeting 4 hours. Show up and feel free to ask questions. Don will share how they created thumbnails with....IIIF? This will be the third one.
Danny has it integrated with Islandora 8 right now. Not bad to do.
- don't have to override the spring file
- don't have to deploy all the same things.
- configured stuff passed in via java opts
- running tomcat command as a service
- creates everything just fine
- using State Tokens to help update data
- Milliner is configured to use either fedora 5 or 6. Milliner will make different requests based on which version you configure it for
- Danny tested import/export tooling for fedora 4=>5=>6. Danny found an issue with file names with spaces and once that minor thing is fixed, it seems like it's good to go and works fine.
- no visible differences seen from the Islandora layer
- Having Islandora transition from using Fedora 5 to 6 has been painless.
- Islandora will be Fedora 5 and 6 compliant, people will not have to move to 6 right away.
- Nice work Fedora team!
- media and nodes are still their own object. The Islandora data model in Fedora 6 has not changed.
- Advanced queue has been talked about before. ActiveMQ has no access via drupal, so if something goes wrong, it's hard to debug. If we moved to Advanced Queue we might not need a java stack at all.
- There is now a drupal module named Triplestore Indexer
- this can use Advanced Queue (https://www.drupal.org/project/advancedqueue)
- indexed into triplestore when cron is run or via drush command
- very configurable
- there is an action you can use to cause it to reindex an object(s) into triplestore
- advanced queue has controls to limit how many should be indexed at once (or rather, how long it should spend time processing)
- can you have multiple queues? this way you could have a front end triple store and a back end triplestore.
- Can there be an option added to say "index as seen by anonymous user" (by role)? This way you could limit what data gets indexed. (the idea is perhaps you have a front-end triplestore and you don't want all the data to be available to anon users)
- maybe via namespaces in one blazegraph instance
- some might want to have two blazegraphs - one for internal useage and one for external.
- how do failed requests get re-queued? In admin/content - there is a way on that page. Can the report page have a feature on it to trigger re-start or re-queue. The restarting logic is in the advanced queue module, so it's not straightforward to put it on the page listing results.
- perhaps add any use cases you have to here: https://www.drupal.org/project/advancedqueue
- how scalable is Advanced Queue compared to ActiveMQ? It will be good to see how they compare, performance-wise. Chances are they will compare pretty well.
- using this as a template, we could pull in more of the queues from Active MQ
- we need to make sure this doesn't have this grind to a halt when ingesting a lot of data.
- some have dedicated Drupal for ingest, then have separate Drupal for users (operating on the same data)
- indexing can happen right away by sending to entity action hook, but this hook method doesn't scale.
- we would need some service to keep triggering processing of the queue.
- https://github.com/Islandora/documentation/issues/1744 -- Danny knows how to fix, sounds like some config is needed.
- https://github.com/Islandora/islandora_defaults/pull/46/
- this PR is trying to make it more complete
- there are a lot of deleted files
- Danny will ask Rosie about this
Bethany asked a question about using features vs. config-import/export to manage config changes:
- config-import/export better for site maintenance
- features - now mostly for providing new features, versus maintaining configuration now.
- features - putting features.yml file in module to help pull in new things w/o reinstalling.
- features assumes you are clobbering everything
- way easier to use site config-import/export
Adam noted that optional configuration gets installed when you install another language.
Another option for backing up config: https://www.drupal.org/project/backup_migrate Get a SQL dump w/o having to use the CLI.
You may be looking for the islandora-community wiki · new to islandora? · community calendar · interest groups · roadmap