Bug 1579986 - LDAP user can log into one region and not the other
Summary: LDAP user can log into one region and not the other
Keywords:
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: Red Hat CloudForms Management Engine
Classification: Red Hat
Component: Appliance
Version: 5.8.0
Hardware: Unspecified
OS: Unspecified
high
high
Target Milestone: GA
: 5.8.5
Assignee: Joe Vlcek
QA Contact: Matt Pusateri
URL:
Whiteboard: auth:miqldap:ad
Depends On:
Blocks: 1572700
TreeView+ depends on / blocked
 
Reported: 2018-05-18 22:12 UTC by Tuan
Modified: 2021-12-10 16:11 UTC (History)
12 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-08-07 15:43:52 UTC
Category: Bug
Cloudforms Team: CFME Core
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Comment 3 Joe Vlcek 2018-05-22 18:48:01 UTC
Tuan,

I have written a blog post [1] describing how to diagnose and what to report regarding
authentication issues.

There is a section titled Reporting Authentication Issues [2]

Please have the customer provide the information requested in the section titled:
"Capturing Authentication Data From ManageIQ DB" [3] for both the succeeding and
failing appliance. That being the output from the following commands:

/var/www/miq/vmdb/bin/rails runner 'puts Settings.authentication'
/var/www/miq/vmdb/bin/rails runner 'puts MiqGroup.pluck(:description)'
/var/www/miq/vmdb/bin/rails runner 'puts User.pluck(:userid)'

Also please ask the customer to go through the section titled: "Confirming LDAP Configuration" [4] and provide the output from the example ldapsearch commands for
both the failing and succeeding appliances. That being the output from the following commands:

ldapsearch -x -H ldap://ldaphost.example.com:389  -b "dc=example,dc=com"  -D "cn=Manager,dc=example,dc=com" -w password

ldapsearch -x -H ldap://ldaphost.example.com:389  -b "dc=example,dc=com"

ldapsearch -x -H ldaps://ldaphost.example.com:636  -b "dc=example,dc=com"

This information is critical in allowing me to help diagnose the issue.

Thank you. JoeV
















[1] http://manageiq.org/blog/2018/01/troubleshooting-auth/
[2] http://manageiq.org/blog/2018/01/troubleshooting-auth/#reporting-authentication-issues
[3] http://manageiq.org/blog/2018/01/troubleshooting-auth/#capturing-authentication-data-from-manageiq-db
[4] http://manageiq.org/blog/2018/01/troubleshooting-auth/#confirming-ldap-configuration

Comment 7 Joe Vlcek 2018-05-24 18:01:16 UTC
Tuan,

Please extend my thanks to the customer for all the help and patience with this
issue.

After reviewing the provided data I am fairly sure I realize what is happening.

In the 4.2 appliance there was a bug that was fixed in the newer versions. 

Groups should be manually created in every region but the bug was that groups would match across regions. So groups only had to be manually created in one region. The fix plugged this issue by confirming that the user attempting to log in had matching groups manually created in the region they are logging into.

The new functionality checks that groups were manually created in every region you
want those users to log into.

Please test by manually adding the groups for one user to one region that is
failing. 

If this does not resolve the problem, let's try to setup a conference call to diagnose this directly some time Tuesday or later next week.

Comment 9 Joe Vlcek 2018-05-29 13:30:22 UTC
Taun,

Please also ask them to time how long the ldapsearch command takes on both of
the system. They should take about the same time on both.

e.g.:

time ldapsearch -x -H ldaps://ldaphost.example.com:636  -b "dc=example,dc=com"

Thank you. Joe

Comment 13 Joe Vlcek 2018-05-30 13:34:51 UTC
Taun,

I noticed the customer has not updated the case with the output from "time"
on the ldapsearch commands.

Since the last update the customer states that the failures are observed on
both their 4.5 and and 4.2 environments a likely issues could be LDAP performance
related. Upon reexamining the data they have already provided I see they have
"group_memberships_max_depth" set to "10". This deep of a group membership search
could produce performance issues.

I also noticed they have configured two LDAP servers. It would be best if they
could do some experimenting with using only one at a time to eliminate the
possibility that one server is having performance issues.

Please ask them to provide the output from a command similar to the one below
run on both regions and to both of their LDAP servers:

time ldapsearch -x -LLL -H ldaps://<my ldap server>:636  -b "<my searchbase>"  -D "<my binddn?" -w <my password> -s sub "(userPrincipalName=<my user>@<my user suffix>)" '*' memberof


Please pass this comment on "as is" directly to the customers.

I am also available for a video conf call if needed.

JoeV

Comment 21 Joe Vlcek 2018-06-04 09:57:28 UTC
Tuan,

Please also ask Ben to provide the debugging data as described in comment 17
https://bugzilla.redhat.com/show_bug.cgi?id=1579986#c17

To enable debugging for Authentication please set `:level:` under section `:log:` in the `Configuration/Settings/Advanced` to `debug` the tail the evm.log file while attempting to log in with the failing user ID.

You should see line with:

MiqLdap#...
and
Authenticator::Ldap#...

Thank you, JoeV

Comment 25 Joe Vlcek 2018-06-18 18:02:43 UTC
The customer reports things work fine for some users but users with many more groups are seeing various failures. This indicates to me that the issue is very likely routed in the customers environment. I have offered many times to join a conf call to help diagnose and root cause the issue but the customers seems reluctant to do that.

II propose this case be downgraded or closed.

JoeV

Comment 32 Joe Vlcek 2018-08-07 15:43:52 UTC
Tuan,

This BZ has been in "Waiting on Customer" status for many weeks now.

The description of the problem has also changed during the course of time
this bug has been open, making it hard to understand what is wrong.

I'm going to mark it closed/insufficient_data.

If the customer wishes to reopen please ask them if we can get on a video
conference to see what is failing and diagnose the issue.

Thank you. JoeV

Comment 34 Joe Vlcek 2018-08-20 12:42:00 UTC
Closed. Canceling NEEDINFO


Note You need to log in before you can comment on or make changes to this bug.