Grouper Users - Open Discussion List

["GW Brown, Information Systems and Computing" <>]

Hi Shilen,

Some responses and questions below.

Regards,

Gary

--On 12 December 2006 15:17 -0500 Shilen Patel 
<>
 wrote:

> Hello,
>
> I have Grouper 1.1 installed with about 72,000 groups and 460,000
> members.  After adding all these groups and members, I've noticed
> performance issues like the following:
>
>
> 1.  In the Grouper UI, using the "Manage Groups" link stalls.  However, I
> have no problems using "All Groups".
Is 'Create Groups' similarly slow? Is searching slow?

Both options call:
member.hasAdmin()
member.hasUpdate()

Manage Groups also calls
member.hasStem()
member.hasCreate()

These calls are made so that the UI can filter out stems which don't have 
manageable groups. It would be possible to come up with a configuration 
setting which would avoid these calls - and so should perform as 'All 
Groups' does, however, you would be able to browse down dead ends.

blair: The UI would work OK if it were possible to ask:
does Stem S have any descendant Groups where Subject subj has a privilege 
from [update, admin,...]. This might not be straightforward to determine, 
but would avoid the member.has calls.

> 2.  Since the ALL subject is granted read and view privileges to all of
> the groups, method calls like Member.hasRead() and Member.hasView() can
> take hours.
Granting of read and view privileges to GrouperAll is optional and the 
default privileges granted configurable, but if it is appropriate to your 
scenario then we have to figure out what is reasonable behaviour when the 
number of groups returned is in the 1000's
> 3.  I have an instance where a Group X has admin privileges to about
> 20,000 groups.   It takes about an hour to add a member to Group X
> because the new member is indirectly given admin privileges to the 20,000
> groups.
>
>
> Have issues like these been addressed?  Will they be resolved in a future
> version of Grouper?  Any tips on how I can perform these operations
> faster?
As we get feedback from sites implementing Grouper in real world situations 
and with large repositories we will attempt to resolve the issues that come 
up. Those that are a barrier to uptake of the product are likely to be 
dealt with first. There is fairly limited experience of working with large 
repositories, and how Grouper and the UI perform may well vary according to 
how repositories are structured.

Would you be able to provide an overview of how your repository is 
structured? i.e. how many top level stems are there and how deep does the 
hierarchy go?

> Thanks,
>
> -- Shilen



----------------------
GW Brown, Information Systems and Computing