Cloned from Pagure issue: https://pagure.io/SSSD/sssd/issue/3745
- Created at 2018-05-22 10:43:08 by gerases
- Closed at 2018-05-22 12:05:53 as Invalid
- Assigned to nobody
Hi,
I've got a mystery problem with an IPA user (let's call it user A) who doesn't have a password and does not belong to an HBAC rule. This is a role account for some automation I would like to use.
I log in with that user using ssh and GSSAPI based on the user's keytab. Setting the debug level to 0x3ff0 shows that after checking the user's groups, sssd just prints a message about opening a PAM session. No messages about why access was granted.
When I log in as another user that does belong to an HBAC rule, I see messages like this:
Access granted by HBAC rule [SOME_RULE_NAME]
What makes it possible for user A to be granted ssh access without an HBAC rule?
Comments
Comment from jhrozek at 2018-05-22 11:01:33
On 22 May 2018, at 10:43, Sergei Gerasenko pagure@pagure.io wrote:
=20
=20
gerases reported a new issue against the project: sssd that you are =
following:
``
Hi,
=20
I've got a mystery problem with an IPA user (let's call it user A) =
who doesn't have a password and does not belong to an HBAC rule. This =
is a role account for some automation I would like to use.
=20
I log in with that user using ssh and GSSAPI based on the user's =
keytab. Setting the debug level to 0x3ff0 shows that after checking =
the user's groups, sssd just prints a message about opening a PAM =
session. No messages about why he was granted.
=20
When I log in as another user that does belong to an HBAC rule, I see =
messages like this:
=20
`Access granted by HBAC rule [SOME_RULE_NAME]`
=20
How can I see why the user A was granted ssh access without an HBAC =
rule?
If you set the debug level to a higher level you will see all the rules =
evaluated in detail. (setting the debug level to 10 will give you maybe =
even more info than you hoped for =E2=80=A6)
``
=20
To reply, visit the link below or just reply to this email
https://pagure.io/SSSD/sssd/issue/3745
Comment from gerases at 2018-05-22 11:18:07
I've set it to 10 and I do see a lot more output, but still no messages about which HBAC rule allowed the user access. It looks like hbac is just skipped? For other users I see: hbac_rule_element_debug_print lines followed by pam_print_data. For this user, I just see pam_print_data at the end of the login. Nothing about HBAC.
Comment from sbose at 2018-05-22 11:28:01
To make HBAC rules work the service, sshd in this case, must call the needed PAM calls.
Please check if 'UsePAM' is set to 'yes' in /etc/ssh/sshd_config and that pam_sss is called in the 'account' part of sshd PAM configuration, e.g. /etc/pam.d/sshd.
Comment from gerases at 2018-05-22 11:30:37
UsePAM is set to yes and pam_sss is used in the account section of pam configs. Note, HBAC rules are consulted for all other users I've tried.
Comment from sbose at 2018-05-22 11:43:47
Is there by chance a local user with the same name which shortcuts the PAM checks?
Comment from gerases at 2018-05-22 11:44:06
I found the reason!!!
The account line of the pam file has a sufficient rule based on the uid range. And so HBAC logic was completely skipped for that user since its UID fell in the range.
Mystery solved. Thank you, @sbose, for clueing me in on the fact that HBAC logic is turned on in the account portion of PAM.
Comment from gerases at 2018-05-22 11:46:38
Woot, woot! I can now go to bed :)
Comment from sbose at 2018-05-22 12:05:19
:), good night, I'll close the ticket.
Comment from sbose at 2018-05-22 12:05:54
Metadata Update from @sbose:
- Issue close_status updated to: Invalid
- Issue status updated to: Closed (was: Open)
Cloned from Pagure issue: https://pagure.io/SSSD/sssd/issue/3745
Hi,
I've got a mystery problem with an IPA user (let's call it user A) who doesn't have a password and does not belong to an HBAC rule. This is a role account for some automation I would like to use.
I log in with that user using ssh and GSSAPI based on the user's keytab. Setting the debug level to
0x3ff0shows that after checking the user's groups, sssd just prints a message about opening a PAM session. No messages about why access was granted.When I log in as another user that does belong to an HBAC rule, I see messages like this:
Access granted by HBAC rule [SOME_RULE_NAME]What makes it possible for user A to be granted ssh access without an HBAC rule?
Comments
Comment from jhrozek at 2018-05-22 11:01:33
If you set the debug level to a higher level you will see all the rules =
evaluated in detail. (setting the debug level to 10 will give you maybe =
even more info than you hoped for =E2=80=A6)
Comment from gerases at 2018-05-22 11:18:07
I've set it to 10 and I do see a lot more output, but still no messages about which HBAC rule allowed the user access. It looks like hbac is just skipped? For other users I see:
hbac_rule_element_debug_printlines followed bypam_print_data. For this user, I just seepam_print_dataat the end of the login. Nothing about HBAC.Comment from sbose at 2018-05-22 11:28:01
To make HBAC rules work the service, sshd in this case, must call the needed PAM calls.
Please check if 'UsePAM' is set to 'yes' in /etc/ssh/sshd_config and that pam_sss is called in the 'account' part of sshd PAM configuration, e.g. /etc/pam.d/sshd.
Comment from gerases at 2018-05-22 11:30:37
UsePAMis set toyesandpam_sssis used in the account section of pam configs. Note, HBAC rules are consulted for all other users I've tried.Comment from sbose at 2018-05-22 11:43:47
Is there by chance a local user with the same name which shortcuts the PAM checks?
Comment from gerases at 2018-05-22 11:44:06
I found the reason!!!
The account line of the pam file has a
sufficientrule based on the uid range. And so HBAC logic was completely skipped for that user since its UID fell in the range.Mystery solved. Thank you, @sbose, for clueing me in on the fact that HBAC logic is turned on in the account portion of PAM.
Comment from gerases at 2018-05-22 11:46:38
Woot, woot! I can now go to bed :)
Comment from sbose at 2018-05-22 12:05:19
:), good night, I'll close the ticket.
Comment from sbose at 2018-05-22 12:05:54
Metadata Update from @sbose: