2025-02-28 Hyrax Fedora 6 Working Group
Date
Feb 28, 2025@9am Eastern
Participants
@Arran Griffith
@Ben Pennell
@Bradley Watson (absent)
@Collin Brittle (absent)
@Dan Field
@Daniel Pierce
@Emily Porter
@Heather Greer Klein
@Jon Dunn - I’ll be late due to another meeting
@Juliet Hardesty
@Kate Dohe
@Nicholas Mark Homenda
@Scott Prater
@Randall Floyd
@Rebekah Kati
@Tom Wrobel
indicates note taker
Discussion topics
Item |
|---|
Welcome |
Updates:
|
What work remains?
|
Wrap Up Next meeting:
Next note taker: Tom Wrobel |
Notes
Arran has made some updates to the wiki about our general status and dates.
Hyrax release:
Nick Homenda: concluding a maintenance sprint, goal is to have a release candidate and update hosted versions of Nurax. Fedora 6 is well supported, especially compared to the Fedora 4 configuration with Hyrax and Valkyrie: ActiveFedora and Wings still have some issues.
Will this be an RC or a full release? Plan is to deploy the RC to Nurax, and then address the fixes for Fedora 4 support. The full 5.1 release should happen in the next 2 weeks potentially.
Nurax status:
Daniel has been working on getting the Fedora 6 Nurax instance running. Terraform setup has needed some work. This will be S3-backed for the OCFL root. The group notes that this is a very common scenario for most Fedora 6 adopters. The new Fedora 6 instance will be F6.nurax.samvera.org
Discussed that we may want to do QA on Nurax before we request broader community testing with the new instance. Note: the Hyrax Maintenance WG will also be doing their normal QA process.
Performance testing:
If we request additional testing, may want to make a broader community-focused version of the existing QA test plan. The worksheet used by the HMWG is extremely comprehensive.
Sample data for performance testing: to discuss in a future meeting. Do we need a new version of this group’s performance testing worksheet? May want to allow users to only selectively test certain cases, depending on what content they have. We don’t normally host a significant volume of sample data in our Nurax instances, and are not sure if storage costs will be a concern. We could also potentially time box how long the data is retained, if storage cost is an issue. We also may need to consider other AWS costs. Nurax uses ECS (not EC2). This is running under Samvera’s AWS account, so some cost monitoring may be needed.
Emory migration updates:
Ingests running since 1/30 (4 weeks) - 20% complete for production migration.
~7K works; ~56K files ingested as of 2/28
450156 Fedora resources
Not using pairtree
A batch of 1000 works and 8000 files takes ~23 hours
Encountering some issues related to Valkyrized Bulkrax. Individual FileSet entries in a CSV import are not yet supported, so we have had to implement some local workarounds to support our ingest needs. The downside is that error reporting is not very granular, and we see few failed jobs in sidekiq which means a lot of manual QA based on the error reporting we can see in the Bulkrax importer statistics. Errors typically include FrozenError and Valkyrie::Persistence::StaleObjectError. Ben mentioned having seen FrozenErrors in the past as well: FrozenError during concurrent activity in 3.2.11 · Issue #445 · ruby-rdf/rdf
We are also encountering issues with the CreateRelationshipsJob failing and are trying to gather more information/logging to diagnose the behavior. Fortunately, we are still able to use the Hyrax UI to bulk-add works to the collection we are ingesting to, but this can be time consuming.