[openstack-dev] [gate] broken by pyeclib 1.0.9 release

Sean Dague sean at dague.net
Fri Sep 11 11:12:45 UTC 2015


On 09/10/2015 03:45 PM, Robert Collins wrote:
> On 11 September 2015 at 07:23, Robert Collins <robertc at robertcollins.net> wrote:
>> Note that master is pinned:
>>
>> commit aca1a74909d7a2841cd9805b7f57c867a1f74b73
>> Author: Tushar Gohad <tushar.gohad at intel.com>
>> Date:   Tue Aug 18 07:55:18 2015 +0000
>>
>>     Restrict PyECLib version to 1.0.7
>>
>>     v1.0.9 rev of PyECLib replaces Jerasure with a native EC
>>     implementation (liberasurecode_rs_vand) as the default
>>     EC scheme.  Going forward, Jerasure will not be bundled
>>     with PyPI version of PyECLib as it used to be, until
>>     v1.0.7.
>>
>>     This is an interim change to Global/Swift requirements
>>     until we get v1.0.9 PyECLib released and included in
>>     global-requirements and ready patches that change Swift
>>     default ec_type (for doc, config samples and unit tests)
>>     from "jerasure_rs_vand" to "liberasurecode_rs_vand."
>>
>>     Without this change, Swift unit tests will break at gate
>>     as soon as PyECLib v1.0.9 lands on PyPI
>>
>>     * Swift is the only user of PyECLib at the moment
>>
>>     Change-Id: I52180355b95679cbcddd497bbdd9be8e7167a3c7
>>
>>
>> But it appears a matching change was not done to j/k - and the pin
>> hasn't been removed from master.
> 
> I'm going to propose another manual review rule I think: we should not
> permit lower releases to use higher versions of libraries -
> approximately noone tests downgrades of their thing [and while it only
> matters for packages with weird installs / state management things,
> its a glaring hole in our reliability story].

I feel like that's a bad thing to assume of people's systems. What is
the expected behavior of an installer if it discovers installing
OpenStack requires downgrading a library? Halt and catch fire?

It also means we're back to having to pin requirements in stable
branches because that's the only way we can guaruntee this for people.
And that's a thing we specifically wanted to get out of the business of
doing because it let to all kinds of problems.

	-Sean

-- 
Sean Dague
http://dague.net



More information about the OpenStack-dev mailing list