[ 
https://issues.apache.org/jira/browse/RANGER-1768?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16157553#comment-16157553
 ] 

Don Bosco Durai commented on RANGER-1768:
-----------------------------------------

I can't find the JIRA, but there was a discussion on where to use the group 
mapping using Hadoop Common UGI API. E.g. get all the users from LDAP/AD, but 
use Hadoop UGI API get the groups. Not sure what was the resolution on it.

Regardless, my strong +1 on your suggestion. This will grossly simplify our 
group mapping configuration. The only downside is, we can't get the users for 
specific groups.

I also not sure whether we can use the SSSD LDAP configuration to identify a 
user and modify it to do a global search to get the users also from LDAP.

> User Sync: add NSS standard user/group resolver mechanism to transparently 
> support all Linux OS level identity management systems
> ---------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: RANGER-1768
>                 URL: https://issues.apache.org/jira/browse/RANGER-1768
>             Project: Ranger
>          Issue Type: New Feature
>          Components: usersync
>    Affects Versions: 0.7.0
>         Environment: HDP 2.6
>            Reporter: Hari Sekhon
>
> Feature Request to add UserSync support for the standard Linux NSS user/group 
> resolver mechanism to allow offloading user/group integration to the standard 
> OS tools like SSSD.
> This will allow Ranger to sync users and groups from the Linux OS integration 
> layer using the standard user/group resolver modules which will cover all 
> possible mechanisms which can include anything that the widely used SSSD can 
> do including both local and LDAP users (which would obsolete having to 
> configure LDAP manually in Ranger as it would be transparent regardless of 
> whether using Active Directory, Redhat IPA, OpenLDAP it would require no 
> different schema configuration in Ranger etc) and it also allows more 
> flexibility as the integration then becomes the more widely used standard 
> Linux mechanisms, you can even mix different identity mechanisms through this 
> one usersync method, including local accounts and AD / LDAP accounts if 
> needed (some clients have asked for this).
> This is more similar to what Hadoop does, just ask the OS, and is much more 
> flexible, simpler to configure as it's transparent to Ranger once it switches 
> to just doing the NSS lookup, rather than doing its own separate extra LDAP 
> configuration integration directly and ending with up with issues like 
> RANGER-1735 group nesting problems when SSSD solved that back in 2011. 
> Although this group nesting problem is severe enough to likely be fixed soon 
> (it affects customers I'm representing right now too), the point remains that 
> offloading the integration to NSS is by definition more robust, feature 
> complete and more widely tested across many other applications that leverage 
> it.
> This is also a Redhat recommendation, see:
> http://rhelblog.redhat.com/2016/04/26/why-use-sssd-instead-of-a-direct-ldap-configuration-for-applications/



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to