On 01/19/2016 11:52 AM, Pavel Březina wrote:
On 01/19/2016 10:41 AM, Pavel Březina wrote:
> On 01/19/2016 10:37 AM, Pavel Březina wrote:
>> On 01/18/2016 11:30 PM, Michal Židek wrote:
>>> Hi!
>>>
>>> see attached simple patch for ticket:
>>>
https://fedorahosted.org/sssd/ticket/2912
>>>
>>> The first patch is for master and 1.13
>>> the second for 1.12.
>>>
>>> It seems like the decision to ignore cache
>>> validity when background refresh is enabled
>>> was not a good one and it was not possible
>>> to fetch new netgroup entry even if the
>>> old was invalidated with sss_cache.
>>>
>>> Michal
>>
>> Nack.
>>
>> The decision was good, but the patch was wrong. The point was to disable
>> midpoint refresh if background refresh is on. The check should be moved
>> to this area in check_cache():
>>
>> if (ret == EAGAIN) {
>> /* No callback required
>> * This was an out-of-band update. We'll return EOK
>> * so the calling function can return the cached entry
>> * immediately.
>> */
>>
>> something like: if (ret == EAGAIN && is_refreshed_on_bg()) then ret =
>> EOK (just return)
>>
>> Also please extend is_refreshed_on_bg() to return true for users and
>> groups since it is already supported.
>
> Ah, I just saw your second patch that does exactly this thing. I guess
> you can ignore this nack then, I'll review both patches in more detail
> later.
Ack to this patch.
This patch will be merged with another patch in different thread.
So, no need to push the one from this thread.
Michal