Draft Minutes: Grouper Call of Wed. Dec. 3, 2014
Tom Barton, U. Chicago, chair
Chris Hyzer, U. Penn
Shilen Patel, Duke
Jim Fox, U. Washington
Misagh Moayyed, Unicon
Emily Eisbruch, Internet2
New Action Items
[AI] (Chris) create a list of updates to the new UI for Grouper 2.3
[AI] (Jim) will do the 2.1 to 2.2 upgrade again and log times.
[AI] (Shilen) explore using threads in the upgrade process to reduce time required. Will consult with Chris and what was done on the Loader.
Carry Over Action Items
[AI] (Emily) look into providing better, easier to find, documentation on deployment experiences with Grouper and AD, Grouper and Google Groups, etc. perhaps using Confluence tags (not done yet as of Dec 3, 2014. In the meantime, thanks to all the campuses who have provided updates to their Grouper Adopter Sketches.)
[AI] (Shilen) create a Grouper training video on the new Grouper
[AI] (Tom) follow up on Pen testing of Grouper 2.2
Grouper v2.2.1 early field experience
Have heard on the list of small issues, nothing major so far in terms of feedback from the field.
Some mentioned the upgrade takes a long time.
Shilen: Upgrading from 2.2.0 to 2.2.1 does take time when using mysql it seems. It is not a major upgrade and views/constraints don’t need to be recreated (one site reported recreating views/constraints took a long time). So 2.2.2 will have a fix for that.
Upgrading from 2.1 to 2.2 is a major upgrade and does take more time.
Can some of the generating of new tables be done without downtime?
Shilen: yes some can be done in parallel, though there is a risk of some tables getting out of date.
We gave Brown some pointers on reducing the time needed for upgrade.
We need to manage expectations about Grouper upgrades, need to document/ explain that the upgrade to Grouper 2.2 is especially time consuming.
AI (Jim) will do the 2.1 to 2.2 upgrade again and log times.
Data from Duke? Shilen will look for the data, not sure if he still has it.
Can we use threads for some of the upgrade work? Using threads helped for the loader upgrade.
Shilen: yes, we could use multiple threads.
AI (Shilen) explore using threads in the upgrade process to reduce time required. Will consult with Chris and what was done on the Loader.
Shilen; how do we know when it’s OK, there is a sensible amount of time for the upgrade?
Answer: we will do our best
Patching for Grouper
Chris has been working on patching:
Patches are applied:
- When you install Grouper with the installer
- When you upgrade Grouper with the installer
- When you run the installer in patch mode
if some patches can be done automatically it will be easier to maintain the Grouper install. More use of scripts. Duke uses Docker environment, all scripted.
Feedback: this plan for patching looks good
SCIM, interfaces, & IETF
Will discuss PSP at a future call when DaveL is on the call
First pass at what's in v2.3. Cf. Roadmap.
improve folder privileges
improve Loader, make it more available and reliable
Tom: What’s the right way to deal with unresolvables?
Chris: Shilen’s enhancement to the member table to add NET ID (Subject ID) will help this issue.
Best approach: Try to resolve it internally. If unresolvables are below a certain threshold (configurable) then ignore it, or flag it with an email.
Right now any unresolvable causes fail
Jim: I err on the side of leaving unresolvable people in a group. It’s safer
Sometimes people change their Subject ID
Could have an option for Grouper to leave unresolvables alone
Also for the Grouper 2.3 release - finish the new UI
the new patching approach should make it easier to handle small issues, so Chris can focus on Grouper 2.3 development.
Do we need to priortize items for finishing the UI?
First, finish UI items we did not finish for the 2.2 release
Suggestion: keep the old UIs around/available in Grouper 2.3
AI (Chris) create a list of updates to the new UI for Grouper 2.3
Also for Grouper 2.3 - PSP - DaveL will work on this
Add remaining attribute/permission operations to WS (on roadmap but not much demand for this)
Improve GSH - build on what Bill Thompson has done here, using groovy?
Misagh - Unicon has an interest in the GSH issue
Tom: What requirements should we have for GSH? Is there a model to emulate?
Misagh: there are existing models and frameworks for shells. Spring, for example. Docker has an intelligent shell. The shell is wrap around the Grouper API. The Grouper API is not necessarily all that friendly. There is a lot to wire. Would be good to have a unified language around the common operations, to be embedded into the shell. So you don’t have to call the API. You write something on top that is more friendly.
This is a larger conversation. To be continued on a future call.
Next Grouper Call: Wed. Dec.18, 2014 at noon ET