Yeah,

When 'kinit -k' is failing, you have really fundamental Kerberos failures going on.  It's not even involving sssd at that point.  Usually, 
1. your entries in /etc/krb5.keytab file are stale or wrong, or 
2. it can't find or access your AD DCs or 
3. other DNS problems.

You can run 'kinit -k' in debug mode (export KRB5_TRACE=/tmp/krb5.out) and view that debug file to see more specifics about the specific failure.

For instance, we have two Linux VMs in a cloud provider now where the AD integration is misbehaving.  kinit -k is failing.  From the debug log of 'kinit -k', we can tell it's the DNS discovers of the AD DCs that's failing.  Turns out, it's a stealth MTU problem.

I believe until recently, sssd didn't support storing its kerberos creds in the KCM, so your particular error condition is not one I've run across yet.

Glad to hear your problem is resolved.

Spike

On Fri, Nov 19, 2021 at 5:55 AM Aron Kelemen Szabo <aron.kelemen@stralfors.se> wrote:

Hi again!

 

I think I have found the problem! 😊

 

kinit -k returned the following error:

kinit: Connection refused while getting default ccache

…while kvno -S host TEST0003 returned the following error:

kvno: Connection refused while opening ccache

 

A bit of googling on these error messages revealed that the culprit might be the KCM-service not running... When googling a bit on the error messages thrown by kinit and kvno and issuing the following commands revealed that this was indeed the case.

systemctl status sssd-kcm.socket

systemctl status sssd-kcm.service

 

I came across this error report which among other things recommends upgrading the packages: https://bugzilla.redhat.com/show_bug.cgi?id=1716981

 

So I run a “yum update” and re-joined the host to the realm, and now the AD-logons seem to be working fine! Now I “only” need to find out the very happening that rendered KCM to fail. :-)

 

Thanks for all your input again!

 

Best regards,

Áron

 

 

From: Spike White <spikewhitetx@gmail.com>
Sent: den 18 november 2021 18:46
To: End-user discussions about the System Security Services Daemon <sssd-users@lists.fedorahosted.org>
Subject: [SSSD-users] Re: SSSD and Kerberos-related problems during joining a RHEL8.4-host to AD

 

Aron,

 

Several things.    Some backgroun -- in our company, we have thousands of OL8.x and hundreds of RHEL 8.x Linux servers directly AD integrated to our corp AD domain.

 

I compared our sssd.conf with yours.  I think you want to add the 'ifp' service for *L8.  It's the infopipe service.  Used by support utilities such as sssctl domain-list, etc.

 

I thought you had to have a sssd.conf stanza for each service you enable.  For instance, we have this:

 

[nss]
debug_level = 0x0100
#debug_level = 9
filter_groups = root
filter_users = root

[pam]
pam_verbosity = 3
#debug_level = 9
offline_credentials_expiration = 3

[ifp]

 

Because we have 

[sssd]

...

services = nss,pam,ifp

 

 

From your ldap_child.log, it looks like your SASL bind and then the LDAP query is working.  Which is surprising to me.  We set up

 

ldap_sasl_authid = host/<fqdn>@<ad domain>

 

in our sssd.conf file.  But I'm guessing if that's not explicitly set, it uses HOSTNAME$@REALM.  At least that's what it appears from your ldap_child.log. It appears to use:  TEST0003$@ourlab.se

 

Several things to try to narrow down where the failure is occuring.  

1. try 'kinit -k'  on the command line.   That uses the first entry in /etc/krb5.keytab file to attempt to authenticate as this machine account.  It's not a perfect test, since it's acquiring a TGT ticket instead of a service ticket.

2. If that succeeds, try kvno -S host TEST0003

     That should report the KVNO of your machine account credentials, as stored in AD.  This is a better test, as you're acquiring a service ticket here.  (Like sssd does).

3. Try 'adcli testjoin -D ourlab.se -v'.  This tests the sssd connectivity, similar to how sssd does it.

 

if all that works, your problem is not with LDAP or your SASL binding.  So then,

 

4. try 'getent passwd aron.kelemen@ourlab.se'.  That should return the entry for this user.  If this fails, possibly your /etc/nsswitch.conf file isn't set up right or maybe your attribute mapping in AD isn't agreeing with your sssd.conf setting.  (We use the MS-supported RFC2307bis AD schema extension)..

 

5. If that's good, then probably the problem is something in your PAM stack.  Specifically, the auth phase.   In our /etc/pam.d/sshd file, we have:

 

#%PAM-1.0
auth       substack     password-auth
auth       include      postlogin

...

 

and in password-auth, we have:

 

auth        required                                     pam_env.so
# OL7/8 version. Per I/T's stated policy for service & process accounts, lock-out time = 30 mins
auth        required                                     pam_faillock.so preauth silent deny=5 unlock_time=1800
auth        sufficient                                   pam_sss.so forward_pass
auth        sufficient                                   pam_unix.so nullok try_first_pass
auth        requisite                                    pam_succeed_if.so uid >= 1000 quiet_success
auth        required                                     pam_faillock.so authfail deny=5 unlock_time=1800
auth        required                                     pam_deny.so

 

 

Spike

_______________________________________________
sssd-users mailing list -- sssd-users@lists.fedorahosted.org
To unsubscribe send an email to sssd-users-leave@lists.fedorahosted.org
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedorahosted.org/archives/list/sssd-users@lists.fedorahosted.org
Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure