Open Stack

Sean Dague <sean at dague.net> wrote:
> On 09/15/2016 09:20 AM, Roman Podoliaka wrote:
>> Sean,
>>
>> So currently we have a default timeout of 160s in Nova. And
>> specifically for migration tests we set a scaling factor of 2. Let's
>> maybe give 2.5 or 3 a try ( https://review.openstack.org/#/c/370805/ )
>> and make a couple of "rechecks" to see if it helps or not.
> 
> Given the fail rate, lets just go safe and double it to 4. If it's
> really a timeout issue that should move the bell curve to mostly cover
> it. Timeouts are really just supposed to be a backstop to ensure the
> tests don't deadlock forever.
> 
> The incidence rate is at that odd race rate that we'll probably have to
> merge and just see if it goes away.

Neutron also started seeing timeouts in DB migration tests today.
We're going to try a similar timeout scaling factor for these tests.