On 08/12/2015 02:20 PM, Jakub Hrozek wrote:
On Fri, Aug 07, 2015 at 12:22:39PM +0200, Pavel Březina wrote:
> On 07/30/2015 09:52 PM, Jakub Hrozek wrote:
>> On Thu, Jul 30, 2015 at 09:46:11PM +0200, Jakub Hrozek wrote:
>>> Hi,
>>>
>>> the attached patches implement fetching the keytab for one-way trusts on
>>> each sssd restart. This is in order for admin to be able to call service
>>> sssd restart and have fresh keytabs in case the trust was re-established
>>> in the meantime.
>>>
>>> Even though retrieving the keytabs is quite expensive operation,
>>> restarting the sssd instance on the IPA server should be quite rare.
>>
>> Sorry, I shouldn't be sending patches before Coverity results arrive.
>> Attached version fixes error handling in the first patch and fixes an
>> unused variable in the second one.
>
> Hi,
> the code looks good. I just have an idea to move the talloc destructor that
> ensure the temporary file will get unlinked into sss_unique_file.
>
> We can provide a talloc context there and setup a destructor if requested.
> Something like:
>
> sss_unique_file(owner, file)
> if owner != NULL
> talloc_set_destructor
Hi,
please see the attached patches. Since the unique file code is not
totally trivial (even though tested) I will move using the
sss_unique_file() interface in other sssd code into a different patchset
-- I would like to apply these patches to downstream and changing the
mkstemp() calls might be too risky there.
Ack.