Grouper Developers Forum

[Chris Hyzer <>]

I would like to discuss what happens when a change log consumer has an error 
on a record.

When Oracle replication has errors it halts everything (and it queues up) 
until someone fixes it (at least, that is how I have experienced it).

Do we want the same thing to happen here?  If the temp change log daemon has 
an error, no notifications will go out until it is fixed.  If 1 change log 
consumer has an error, then it doesn't affect other change log consumers, but 
no events will go out for that one until it is resolved.

I could picture a system where errors are logged (and notified?) and 
processing continues.  The system that is getting notified might miss a 
group_create, then get some add_members to that group, but maybe it should be 
designed such that an add_member to a group which isn't in the system causes 
it to go back to grouper WS to get the details for the group.

Anyways, should this be configurable?  Is the way it is ok?  Well, Im not 
sure we have time to change this behavior for 1.6, but just thought I would 
throw this out there...

Thanks,
Chris

-----Original Message-----
From: Tom Barton 
[mailto:]
 
Sent: Tuesday, May 04, 2010 6:42 PM
To: Grouper Dev
Subject: [grouper-dev] Grouper design call, Wednesday, 5 May 2010, 1200EDT 
(1700Z)

Wed 5 May 2010, 1200EDT/1700BST/1800CEST, 60 minutes

NOTE: This is an off-week call we held in reserve for last minute 
coordination of the v1.6 release.

+1 734-615-7474 PREFERRED
+1 866-411-0013 (US/Canada only and only if above number costs you more
than 800/866 calls)

PIN: 0121717 (followed by "#")

SIP:
http://edial.internet2.edu/call/0121717
sip:


agenda:

1. administrivia
    . <http://www.internet2.edu/membership/ip.html>
    . approve minutes
    . review AIs
    . agenda bash

2. v1.5.4??

3. I2 debrief

4. v1.6 release coordination

5. else

Tom