[openstack-dev] [nova] v3 integrated tests taking a lot longer?

Matt Riedemann mriedem at linux.vnet.ibm.com
Wed Nov 19 22:58:17 UTC 2014



On 11/19/2014 3:34 PM, Andrew Laski wrote:
>
> On 11/19/2014 04:16 PM, Jay Pipes wrote:
>> On 11/19/2014 04:00 PM, Matt Riedemann wrote:
>>> On 11/19/2014 9:40 AM, Jay Pipes wrote:
>>>> On 11/18/2014 06:48 PM, Matt Riedemann wrote:
>>>>> I just started noticing today that the v3 integrated api samples tests
>>>>> seem to be taking a lot longer than the other non-v3 integrated api
>>>>> samples tests. On my 4 VCPU, 4 GB RAM VM some of those tests are
>>>>> taking
>>>>> anywhere from 15-50+ seconds, while the non-v3 tests are taking less
>>>>> than a second.
>>>>>
>>>>> Has something changed recently in how the v3 API code is processed
>>>>> that
>>>>> might have caused this?  With microversions or jsonschema validation
>>>>> perhaps?
>>>>>
>>>>> I was thinking it was oslo.db 1.1.0 at first since that was a recent
>>>>> update but given the difference in times between v3 and non-v3 api
>>>>> samples tests I'm thinking otherwise.
>>>>
>>>> Heya,
>>>>
>>>> I've been stung in the past by running either tox or run_tests.sh while
>>>> active in a virtualenv. The speed goes from ~2 seconds per API sample
>>>> test to ~15 seconds per API sample test...
>>>>
>>>> Not sure if this is what is causing your problem, but worth a check.
>>>>
>>>> Best,
>>>> -jay
>>>>
>>>> _______________________________________________
>>>> OpenStack-dev mailing list
>>>> OpenStack-dev at lists.openstack.org
>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>
>>>
>>> So...
>>>
>>> Run: 13101 in 21466.460047 sec
>>>
>>> ...
>>>
>>> mriedem at ubuntu:~/git/nova$ du -sh .testrepository/
>>> 1002M    .testrepository/
>>>
>>> I removed that and I'm running again now.
>>>
>>> We remove pyc files on each run, is there any reason why we couldn't
>>> also remove .testrepository on every run?
>>
>> Well, having some history is sometimes useful (for instance when you
>> want to do: ./run_tests.sh -V --failing to execute only the tests that
>> failed during the last run), so I think having a separate flag (-R) to
>> run_tests.sh would be fine.
>
> Testrepository also uses its history of test run times to try to group
> tests so that each thread takes about the same amount of time to run.
>
>>
>> But, then again I just learned that run_tests.sh is apparently
>> deprecated. Shame :(
>>
>> -jay
>>
>> _______________________________________________
>> OpenStack-dev mailing list
>> OpenStack-dev at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>

Turns out that my huge .testrepository is not apparently the issue, so 
I'll press on.

-- 

Thanks,

Matt Riedemann




More information about the OpenStack-dev mailing list