[Freeipa-devel] [PATCH] 0003 Honor the default home directory in user_add

[Rob Crittenden]

Martin Kosek wrote:
> On Wed, 2012-02-08 at 08:22 -0500, Rob Crittenden wrote:
>> Martin Kosek wrote:
>>> On Tue, 2012-02-07 at 16:31 -0500, Rob Crittenden wrote:
>>>> Petr Viktorin wrote:
>>>>> On 02/07/2012 01:52 PM, Petr Viktorin wrote:
>>>>>> Honor the default home directory base when creating a new user. Test
>>>>>> included. I also cleaned up the way home directory was created.
>>>>>>
>>>>>> This patch removes the default from the --homedirectory option, letting
>>>>>> the server fill it in pre_callback. If I'm reading this correctly,
>>>>>> default_from and create_default run on the client-side, so they can't
>>>>>> get to the config without round-tripping to the server.
>>>>>>
>>>>>> https://fedorahosted.org/freeipa/ticket/2332
>>>>>>
>>>>>> Also, I've cleaned up the home directory generation to use
>>>>>> posixpath.join instead of '%s/%s' and ad-hoc cleanup. This should be
>>>>>> more robust. (It will also behave differently if the username starts
>>>>>> with '/' or maybe similar cases of the user asking for trouble.)
>>>>>>
>>>>>> A question: Do we want to use posixpath here, or os.path? Put another
>>>>>> way, should the home directories separated by '\' if the server runs on
>>>>>> Windows?
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Freeipa-devel mailing list
>>>>>> Freeipa-devel at redhat.com
>>>>>> https://www.redhat.com/mailman/listinfo/freeipa-devel
>>>>>
>>>>> Martin told me I need to make two changes: remove autofill along with
>>>>> default_from, and since I have touched the API, update API.txt.
>>>>>
>>>>> Attaching the updated patch.
>>>>
>>>> This works well. I noticed that the default shell has the same problem.
>>>>
>>>> I wonder if we should roll that similar change in or open a separate ticket.
>>>>
>>>> rob
>>>
>>> Hm, default shell works for me:
>>>
>>> # ipa config-mod --defaultshell=/bin/bash
>>> # ipa user-add --first=Foo --last=Bar fbar2
>>> ------------------
>>> Added user "fbar2"
>>> ------------------
>>>     User login: fbar2
>>>     First name: Foo
>>>     Last name: Bar
>>>     Full name: Foo Bar
>>>     Display name: Foo Bar
>>>     Initials: FB
>>>     Home directory: /home/fbar2
>>>     GECOS field: Foo Bar
>>>     Login shell: /bin/bash<<<<   config is honored
>>>     Kerberos principal: fbar2 at IDM.LAB.BOS.REDHAT.COM
>>>     UID: 480800097
>>>     GID: 480800097
>>>     Password: False
>>>     Member of groups: ipausers
>>>     Kerberos keys available: False
>>
>> Odd, I did exactly the same thing and got the wrong shell.
>>
>>> Oh, one more thing that came up to my mind when testing config plugin.
>>> Rob, why do we have config params as optional? We don't expect that the
>>> config attribute is missing in LDAP and IPA crashes in such cases (as in
>>> ticket 2159). IMO they should all be required.
>>
>> So that on a mod you don't have to provide all values. I think we need a
>> non-empty option.
>>
>> rob
>
> mod operation does not require all required options to be passed. You
> can simply update just one (required) attribute, it just must not be set
> to None - which is exactly what we want:
>
> # ipa config-mod --searchrecordslimit=
> ipa: ERROR: 'ipasearchrecordslimit' is required
> # ipa config-mod --searchrecordslimit=150
>    Maximum username length: 32
>    Home directory base: /home
>    Default shell: /bin/bash
>    Default users group: ipausers
>    Default e-mail domain: idm.lab.bos.redhat.com
>    Search time limit: 2
>    Search size limit: 150
>    User search fields: uid,givenname,sn,telephonenumber,ou,title
>    Group search fields: cn,description
>    Enable migration mode: FALSE
>    Certificate Subject base: O=IDM.LAB.BOS.REDHAT.COM
>    Password Expiration Notification (days): 4
>    SELinux user map order: guest_u:s0$xguest_u:s0$user_u:s0-s0:c0.c1023
> $staff_u:s0-s0:c0.c1023$unconfined_u:s0-s0:c0.c1023
>    Default SELinux user: guest_u:s0
>
> You can verify it with the attached testing patch. If you agree, I will
> create a new ticket to do this change and send a proper official patch
> for that.
>
> Martin

ACK. At one time this would cause all options to be prompted when 
executed interactively.

rob