[Building Sakai] LDAP Crashing when more than 2000 users per site

[Andrei Chirila]

Hi Sam,

I think your patch might actually help us solve the problem.

I'll try it out and let you know if it doesn't solve it.
Many thanks for your reply.

Sent from my iPad

With kind regards,

Andrei Chirila
Team Leader

T / +31 (0)356 293972
T / +40 (0)364 403 900 - int. 3227
M / +40 (0)761 210 440
Andrei.Chirila at isdc.eu<mailto:Andrei.Chirila at isdc.eu>

On 12.06.2012, at 10:29, "Sam Ottenhoff" <ottenhoff at longsight.com<mailto:ottenhoff at longsight.com>> wrote:



The idea is that the jldap provider generates quite a heavy LDAP query:
|((uid=111000)(uid=111001)...(uid=113000)). Our Oracle OID LDAP vendor is refusing to run this query so we are looking for alternatives:

So it looks like you are using a custom LDAP provider that implemented getUsers as one batch request?  The default JLDAP provider in Sakai calls to LDAP once per user.


1. We group all users within a site in an LDAP group as well, so the query will be dramatically simplified. But then we will have to maintain users in a site also in the LDAP group. This also means that we have to patch the project.

I'm unclear on this point, but maintaining memberships in multiple places sounds like a pain.



2. We patch the LDAP provider to query LDAP in batches of x. This solution has quite an impact.

Can you provide a link to your patch?  It sounds like you need to implement batching in your requests to not exceed the LDAP server's maximum results limit.

I submitted a patch to the JLDAP provider that uses a configurable maxObjectsToQueryFor:

https://jira.sakaiproject.org/browse/SAK-21105

--Sam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20120612/4d49c601/attachment.html