Attending 

  • Chris Hyzer, Penn, Chair
  • Shilen Patel, Duke
  • Chad Redman, University of North Carolina Chapel Hill
  • Vivek Sachdiva, independent
  • Jeff Williams UNCG
  • Carey Black, the Ohio State University
  •  Emily Eisbruch, Internet2

 

 Administrivia



 

New Action Item from this call

  •   AI Shilen -    Create a JIRA and assign to  Chris Hubing for  adding metadata for registration to Grouper Demo Site Access
  • AI Chris  - look at how to use the UI text for localized error messages


Discussion

Roadmap 

  • Grouper roadmap engagement with the community is being worked on 
  • SteveZ will provide best methods for community outreach around Grouper 3.0
  • Ideas:
    Next 6 months, finish Grouper 2.5, including
    • Provisioners
    • Improved performance
    • Next generation of subject sources
    • Migration of subject sources
  • Next version will be Grouper 3.0
  • Ask community for medium sized tasks needed in Grouper

 

  • Wait for Grouper 3.0 to add new features
    • Example: secure file for report
    • Membership not hitting failsafe limits
  • One month fixing JIRAs
  • Message will be:
    • Grouper 2.5 will be long term support
    • Grouper 2.5 is the supported version
    • Monthly releases of 2.5 for bug fixes, low risk upgrade
    • Please upgrade to it
    • Migrate to new subject sources and new  provisioning

  • Decision around migrating to postgres 
    • Better performance  
    • Document how to run postgres for those who dont run it
    • Performance benchmarks 
  • Database is using too much resource
  • We evolved the Grouper database with many layers
  • Use less memory in the tomcats
  • Change existing tables or make shadow tables


  • Same challenge with indexes, if we start over with structure it can become more efficient
  • Multiplicity of queries due to need to run tests
  • We can become more effective
  • Cache things and clear out cache for testing
  • Make production more efficient
  • Are we supporting batching efficiently?
  • Make sure things are batchable
  • Memberships and attribute assignments are most important
  • Like Shilen improved the changelog, using batching


  • Do we want to keep hibernate?
  • Adds overhead
  • Does Clobs and blobs seamlessly
  • Does paging well
  • That may not be an issue if we go to Postres
  • May not want to move away from Hibernate


  • Move away from ehcache?
  • Be more mindful of when we are using ehcache?
  • Privileges , using a lot of chaining code
  • Web Service w JSON
  • Soap XML can be off by default
  • Grouper Client is using XML , can switch to JSON
  • Get rid of old libraries with issues
  • Consider items that would make a big improvement
  • question: what would replace hibernate with?
  • Look for something lighter weight
  • Need to research this
  • Nested groups
  • Query planner issues
  • Grouper fields table is because we used to have custom lists
  • Memberships table may be used for too many items
  • Split things out into single purpose tables may help with performance 
  • UI issues, so many pixels wide 
  • Challenging for visualization
  • Upgrade all libraries… should be part of build/release cycle, should be automated
  • Shilen: Some UI improvements may be possible prior to Grouper 3.0
  • Table redesign, the views would still work as is?
  • Yes , or perhaps have some Grouper v3 views that are more efficient
  • Handling of UUID
  • Keep legacy working and also evolve
  • Provide a way to migrate over time
  • One database only could be an issue for some sites
  • Backups with postgres require some support
  • Matt: Folder privacy is important 
  • Suggestion to have folder privilege
  • Set of hidden groups where each group maps to a folder
    • If you add  a folder privilege, leave opportunity to use loaders
    • We will need to delve into the best approach
  • JasperReports has stand alone web server now
  • Present an outward facing view with priv based logic?
  • Add a reporting engine bolt on
  • Making all existing audit report things work
  • Documentation improvements
  • Matt: GITLAB, moving away from wiki
  • Helps with version control
  • Jeffrey: migration from Oracle to Postgress could be an issue for some
  • Cosmetic issues with the Grouper UI

 

Current Work

 

Vivek

  • Worked on object types
  • Attributes on stem, more consistent model
  • Deprovisioning and attribute propagation will happen in background 
  • Shilen worked on minimizing number of queries
  • Improving efficiency
  • Changelog consumer controls the process
  • Provisioning is  special case since there are sync tables
  • Request for Chris to review Vivek’s work
  •  
  • Now working on Custom UI
    • The community has not taken advantage of Custom UI yet
    • JSON and attributes on groups is a speed bump
    • Need to migrate how Custom UI works
    • Use Grouper Config, like GSH template works,
    • Model config of custom UI and have a wizard
    • Meta Config into properties file
    • Migration utility to take JSON and migrate to properties

 

Chris

  • Did a build, Maven worked
  • Hope to get build out today
  • 70 Jiras 
  • GSH template fixes
  • Some reporting fixes
  • Manage interface for VPNs
  • Adding more validation to provisioning
  • First level of validation is stuff in JSON and config file

 

Shilen

  • Made performance adjustments
  • LDAP DAO was getting repeated queries, resolved that
  • Code as is now, when you go into UI and set something as provisionable, there is direct assignment, gets propagated, changelog runs as part of provisioner, replicates to group sync table
  • UI will still allow provisionable, but skip setting in attribute framework,
  • Gets set in sync tables
  • Changed to return what propagation should be
  • Makes LDAP full sync test pass
  • Will work on metadata issues and incremental
  • Need to add a column in Grouper sync table for groups for metadata
  • How does that fit into group propagation?
  • Chris: If we keep the member and membership metadata in attributes, since no indirect assignments, then we are OK
  • Shilen will focus on groups, not on membership
  • To add column in Grouper sync group table,
  • Chris has a wiki on DDL changes that will help
  • DDL and Grouper v2.5
  • Shilen: Work again on DN overrides

 

 

Chad

  • Doing Grouper projects for UNC
  • GSH Templates on mock database
  • Used InteliJ so aware of Grouper libraries
  • Worked well, has not gone out to actual environments yet


  • Provisioning work: less straightforward
  • Lining up attributes was an issue
  • Issue on provisioning: multiple types you want to provision, person type or application types. 
  • About 50K subjects
  • Design issue: if you have multiple subject types to provision, all LDAP, you can have person types UID= or application types, CN=, with different subject sources and different checkboxes, how would you look them up, you just have single field,
  • What Logic to use for this 
  • Need to think of logic, perhaps related to subject source
  • Documentation is conceptual, it not explicitly telling you what  “you need this in this field”
    • Name field is DN… must calculate it
  • Hard to know what’s a bug, issue of being an early adopter
  • Issue: have read-only version of LDAP for subject queries
  • Not accessible by the UI
  • Hard to test, see GRP 3402
  • Test config issue
  • Fix is: hard code error messages?
  • For localized error messages, don’t tie them to the UI
  • Should GSH have access to the Jar?
  • Must create fake HTTP Server request
  • AI Chris will look at how to use the UI text for localized error messages
  • Chad’s Moonshot items for Grouper 3.0 and beyond
    • Web services should be beefed up, quirky way of doing REST
    • Consider swagger

 

 

 

Issue Roundup 

 

Jiras in past two weeks

GRP-3383 gsh template error message is misleading (about membership import)

GRP-3382 custom ui gsh exec should throw exception from gsh

GRP-3381 typo in AttributeAssignToAssignmentSave checking type of attribute (multi assign or not)

GRP-3380 improve group/stem report output

GRP-3379  improve GSH template redirect after run

GRP-3378 showEl on gsh templates has issues on hidden field

GRP-3377  group/stem reports on daemon screen show innocuous error

GRP-3376 improve AttestationStemSave

GRP-3375
add manual and intermediate types to visualization legend

GRP-3374
add sql dropdowns to gsh templates based on user

GRP-3373
add useful imports to gsh groovy profile (since compiles are cached) and make them java friendly

GRP-3372
add ability to call a gsh template from another template and consolidate output




 Grouper Emails in past two weeks

 


Grouper wiki updates in past two weeks

 

 

 

 

  • No labels