Versions Compared

Old Version 4

changes.mady.by.user Aaron Birkland

Saved on

compared with

New Version Current

changes.mady.by.user Andrew Woods

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  1. Feature complete
    1. API test suite fails (Indirect Containers, State tokens, etc)
    2. Containment for Mementos
    3. Minimal Search Service UI
    4. Performant rebuild
    5. Side-loading / targeted rebuild
    6. Performant S3 integration (question)(how to measure) 
    7. Object Locking
    8. Session Rollback
  2. Documented performance criteria for 6.0 releasePerformance criteria (question)
    1. Scale
    2. Response times underload
    3. Migration rates
  3. Documentation
    1. Documented API compliance of Beta
    2. Documented missing features of Beta
    3. Documented breaking changes of Beta
    4. Documented upgrade process for Beta

During Beta Testing

  1. Validation tool for F3→F6 migrations
  2. Islandora integration
  3. Samvera integration

Notes

Introductions

  • Folks introduced themselves
  • Purpose of meeting is to keep governance to keep governance and technical groups in sync

...

  • Tech team has been working on performance, metrics.  The dashboard has been a valuable tool
  • Updates to the user interface and API
  • Integration from S3 has been brought in
  • Work on tweaking mementos
  • Work around migration tooling has been an a highlight
    • tooling for migrating from 3,4,5 has been a prerequisite for alpha, all are usable now.
  • Docker containers and automation
  • Work making Fedora cross-database compliant (postgres, mariadb, mysql)
    • Travis runs suite of integration tests across all DBs
    • Putting energy into postgres
    • Postgres is usually the best/fastest/most reliable choice
    • Should we focus intense optimizations to other DBs?
    • Islandora folks might use mariadb more often "by default"
    • Samvera is likely to go postgres
    • --- pause for zoom technical difficulties ---

    • Should there be a recommendation/warning for islandora folks who would tend to use mysql based on Drupal?
      • Difference is not large/horrible; postgres is just slightly faster in general 
  • Validation tool designed, scheduled for January 
    • Bidirectional verification:  Is there anything in fcrepo3 that hasn't been migrated; is there anything in fcrepo6 that didn't come from fcrepo3

...

  • We are extremely close to an alpha
    • Migration tooling is "alpha" ready, a couple outstanding PRs remain
    • Fcrepo6 is mostly feature complete (beta will be fully feature complete)
  • Given the fact that we are mostly feature complete, have migration, and have documentation; are there any barriers to an alpha?
    • What are the criteria from alpha->beta→6.0?  
      • We assume there will be people testing alpha and filing bug reports, providing feedback.  Is this true?  Does this help for when we determine transition from alpha to beta?

      • Andrew:  Think of alpha→beta as a matter of feature completeness.  Beta will be feature complete, and ready to test in institutional environments.  Not sure how much we can rely on early feedback in the alpha phase.
        • Thinking the transition from alpha to beta will be quick; beta to 6.0 is more rigorous.
    • Do we agree with the "Beta Criteria" listed on this wiki?
      • Add Object locking:  two people on the same object.  Right now, second person editing a contended object gets rejected.  Want to fix that
      • Validation tool in some state of beta
      • Feasible to test integration with Samvera or Islandora?  
        • Samvera is diverse.   Hyrax, Avalon, and Hyku have their own needs.  Not so simple to say "hey, can you see it if it works?".  However, there is an initial proposal for a hyrax grant submitted which might have some fedora 6 migration work included in it.  
        • Islandora could do sanity check, but probably not scale check
          • Dustin Slater interested in helping with such a check
        • We want to make sure that nothing is broken as far as APIs staying substantially the same
  • It would be desirable to get a beta out this calendar year.  Feasible to get alpha after next week's sprint.  If we're just talking about feature completeness, a November beta might be possible.  Given validation and integration testing, that might make 2020 harder to hit.  Is that a concern?
    • Scott:  comfortable with alpha now,  Time lag between alpha and beta is not a bad thing, 1Q 2021 is reasonable.
    • David W:  Concern from grant perspective.  It's OK to start at alpha state, but we do need to move these pilots into production.  Planned for beta in 2020, full release early 2021.   There is risk of delaying too long.  Not sure we agree on what Beta is
    • Andrew: Could do DSpace approach of Beta1, Beta2, Beta3, etc... 
  • Anything Any concerns/interest in S3 integration?
    • Fcrepo5 used S3 as binary store (only).  F6 uses it for everything (objects, binaries, manifests, etc).  Need to characterize what that means in real life.

...