On (02/10/14 15:56), Nikolai Kondrashov wrote:
On 10/02/2014 03:19 PM, Michal Židek wrote:
>On 10/02/2014 01:59 PM, Nikolai Kondrashov wrote:
>>On 10/02/2014 02:09 PM, Michal Židek wrote:
>>>On 10/02/2014 10:42 AM, Nikolai Kondrashov wrote:
>>>>On 10/02/2014 12:54 AM, Lukas Slebodnik wrote:
>>>>>On (01/10/14 20:19), Nikolai Kondrashov wrote:
>>>>>>On 10/01/2014 06:33 PM, Jakub Hrozek wrote:
>>>>>>>On Wed, Oct 01, 2014 at 04:58:37PM +0200, Pavel Reichl wrote:
>>>>>>>>
>>>>>>>>On 10/01/2014 04:09 PM, Michal Židek wrote:
>>>>>>>>>Hi,
>>>>>>>>>
>>>>>>>>>see attached simple patch that fixes ticket:
>>>>>>>>>https://fedorahosted.org/sssd/ticket/2453
>>>>>>>>>
>>>>>>>>>Michal
>>>>>>>>>
>>>>>>>>>
>>>>>>>>Patch fixes the problem. ACK
>>>>>>>
>>>>>>>* master: f6e008fe677ffcf8781c1d5154544066ea7ca9d4
>>>>>>
>>>>>>A CI job for this commit has failed on Debian:
>>>>>>http://sssd-ci.duckdns.org/logs/job/1/22/summary.html
>>>>>>
>>>>>>in test-negcache test:
>>>>>>http://sssd-ci.duckdns.org/logs/job/1/22/debian_testing/ci-build-debug/test-negcache.log
>>>>>>
>>>>>>
>>>>>>
>>>>>>Nick
>>>>>Works for me without any problem on clean, minimal, up to date
>>>>>installation of
>>>>>debian-jessie and ubuntu 14.04.
>>>>
>>>>Yes, it was probably a non-reproducible "fluke", but it's
worth being
>>>>aware
>>>>of. I haven't seen it fail ever before.
>>>
>>>It looks like the execution of the test slowed
>>>down (was paused or whatever) and the negative cache expired.
>>>Maybe the VM did not have enough resources from the host and
>>>slowed down execution. Or some other random strangeness.
>>
>>Ah, thank you Michal. It then joins the dyndns test (and some other) in the
>>group of timing-dependent tests. We'll have to do something about these.
>>The
>>VM environment load is not very predictable and it's better not to have
>>randomly-failing tests in CI for whatever reason. Can we somehow
>>increase the
>>time window size for them, perhaps?
>
>Increasing the time window is possible. It is currently set to
>one second in the tests. If it fails in the future again, we
>can double it. But one second should be really a lot for what
>it does so I would keep it as is for now.
Sure, let's see how it goes, thank you.
I see dyndns failures much more often, do you know if we can do anything about
that?
It is a known problem
https://fedorahosted.org/sssd/ticket/2283
Unfortunately, it is in deferred bucket.
Feel free to assign that ticket to yourself.
LS