(besides add the passSyncManagersDNs attribute of course!)

Andy

On Thu, Feb 4, 2021 at 1:13 PM Alfred Victor <[email protected]> wrote:

>
> From:
> https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/linux_domain_identity_authentication_and_policy_guide/migrating_from_a_directory_server_to_ipa
>
> *"Users cannot authenticate to the IdM domain or access IdM resources
> until they have Kerberos hashes."*
>
> To be sure I understand, without the kerberos keys SSH auth to an IPA
> system requiring for instance SSH key+password in sshd_config will fail,
> and 39.1.2.3 as described at link above won't generate the keys
> transparently? Just trying to be sure I understand the scope and what
> actions we need to take to make sure that logins will work, and that the
> necessary keys will get created for a user to not notice whether they are
> on an IPA or an LDAP system, or if this will require further action from us.
>
> Also it sounds like we'll need to use ldapmodify to connect to the IPA
> system to write the password hash. It sounds like I can use the same user
> for this bind that does the migrate-ds currently. This user is currently
> set up with least privilege, has only a special group we have created. Is
> there anything I need to give it specially to be able to write hashes of an
> existing user, or I guess it already can do this if it can do a migrate-ds
> and create users (and their hash values)?
>
> Andy
>
> On Thu, Feb 4, 2021 at 12:38 PM Rob Crittenden <[email protected]>
> wrote:
>
>> Alfred Victor via FreeIPA-users wrote:
>> > Hi Rob and IPA list -
>> >
>> > The alternative is if it is possible to use the sssd method similar to
>> > as described in 39.1.2.3 at the below link to update credentials at IPA
>> > as well when a user resets their password, but I expect that if this is
>> > possible to do with both systems in parallel (OpenLDAP remaining source
>> > of truth until IPA is given the reins), the configuration will be
>> > complex. Hence why I feel we are better off writing this value directly,
>> > or else our IPA systems are going to have a stale credential problem
>> > because migrate-ds will copy the user, or make sure any new user ends up
>> > in IPA, but eventually the user changes password and migrate-ds will
>> > only be concerned that the user already exists and will not update the
>> > credential. Can you please let us know what we'll need to change or do
>> > to write to the password hash directly? I do not plan to store any
>> > credential hashes during this process - nor do I plan to have them
>> > accessible by any means. They will be processed into IPA and that's it.
>> > We already sync public key, group changes to IPA.
>> >
>> > SSSD method:
>> >
>> https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/linux_domain_identity_authentication_and_policy_guide/migrating_from_a_directory_server_to_ipa
>> >
>> > Another option is of course to drop all users and remigrate them, but
>> > this is incompatible with our use case, as we won't have this
>> > opportunity easily.
>>
>> The problem with using a pre-hashed password is that IPA cannot generate
>> new keys so the user would always have to migrate their password using
>> either SSSD or an LDAP bind. How often that would be depends on how
>> often passwords are changing.
>>
>> Syncing in a password change will also trigger an expired password. So
>> if you can grab the unhashed password and sync that then see
>>
>> https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/identity_management_guide/pass-sync
>> section 15.6.3 (ignore everything else, literally).
>>
>> This will allow an unhashed password to be written without marking it as
>> expired.For DN you use the user that binds from your existing password
>> source to do the write.
>>
>> rob
>>
>> > Andy
>> >
>> > On Wed, Feb 3, 2021 at 3:10 PM Alfred Victor <[email protected]
>> > <mailto:[email protected]>> wrote:
>> >
>> >     Hi Rob,
>> >
>> >     We have our system in migration mode. Due to the number and
>> >     diversity of systems we have, we've found we really need to test
>> >     everything extensively before we can cut over and make our IPA
>> >     cluster our source of truth. Keeping a subset of systems on IPA
>> >     during this time allows us to be comfortable switching at some
>> >     future date, given that we know we've had no issues with each subset
>> >     x of all systems y with t duration of production utilization.
>> >
>> >     Andy
>> >
>> >     On Wed, Feb 3, 2021 at 2:08 PM Rob Crittenden <[email protected]
>> >     <mailto:[email protected]>> wrote:
>> >
>> >         Alfred Victor via FreeIPA-users wrote:
>> >         > Hi all,
>> >         >
>> >         > We have a need to set the password hash value directly, is
>> this
>> >         > possible? It does not appear that ipa user-mod will support
>> >         this, and
>> >         > using the API or other methods looks like it will be fraught
>> >         with access
>> >         > control complications.
>> >
>> >         Why do you need to set the hash directly?
>> >
>> >         It is possible to do when adding a new user for migration
>> >         purposes, but
>> >         after that it is generally not allowed.
>> >
>> >         rob
>> >
>> >
>> > _______________________________________________
>> > FreeIPA-users mailing list -- [email protected]
>> > To unsubscribe send an email to
>> [email protected]
>> > 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/[email protected]
>> >
>>
>>
_______________________________________________
FreeIPA-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
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/[email protected]

Reply via email to