[SSSD] remove nisNetgroupTriple or memberNisNetgroup from server bug sssd did not reflect it.

Joe Jin joe.jin at oracle.com
Fri Jan 20 00:21:30 UTC 2012


On 01/20/12 00:17, Stephen Gallagher wrote:
> On Tue, 2012-01-17 at 10:00 +0800, Joe Jin wrote:
>> On 01/16/12 20:40, Stephen Gallagher wrote:
>>> On Mon, 2012-01-16 at 16:48 +0800, Joe Jin wrote:
>>>> On 01/07/12 06:39, Dmitri Pal wrote:
>>>>> Do not have a code in front of me that this sounds like a bug in the
>>>>> error handling of the empty list.
>>>>> It seems that if the list is not empty the cache is reconciled by if the
>>>>> content of the netgroup is empty it chooses a different code path rather
>>>>> than reconciling the content of the group.
>>>>>
>>>>> Stephen, Jakub can you please check my theory?
>>>>
>>>> I tried to call sysdb_remove_attrs() in either sysdb_add_netgroup() or sdap_save_netgroup(),
>>>> when the attr empty, but always return FAILED, any comments?
>>>>
>>>
>>> What was the failure error code? Also, did you use SYSDB_MOD_REPLACE or
>>> SYSDB_MOD_DEL? The latter is probably what you were looking for.
>>>
>>
>> It returned ENOENT(-2) when call sysdb_remove_attrs() in the end of sdap_save_netgroup().
>> In sysdb_remove_attrs(), it using SYSDB_MOD_DEL for ldb_msg_add_empty:
>>
>> 2859         DEBUG(8, ("Removing attribute [%s] from [%s]\n",
>> 2860                   remove_attrs[i], name));
>> 2861         lret = ldb_msg_add_empty(msg, remove_attrs[i],
>> 2862                                  LDB_FLAG_MOD_DELETE, NULL);
>> 2863         if (lret != LDB_SUCCESS) {
>> 2864             ret = sysdb_error_to_errno(lret);
>> 2865             goto done;
>> 2866         }
>>
>> Thanks,
>> Joe
> 
> 
> Sorry Joe, I thought I replied to this earlier, but either I forgot to
> do so or it got lost in the mail.
> 
> Would you mind attaching the patch you were working with? It might be
> easier to track down what's going wrong.

Stephen, as a workaround, I add a empty entry then remove it later, I dont
think it's a good patch at all, so I did not send to the maillist, if you 
interesting it I'll send it to you.

Thanks,
Joe



More information about the sssd-devel mailing list