[openstack-dev] [Neutron] Gate failure

Kevin Benton blak111 at gmail.com
Thu Jan 7 06:38:06 UTC 2016


Ah, that was a bit misleading. We're not leaking any floating IPs. The
create method for them adds them to a list that is cleaned up on teardown.

The issue was that the get_unused_ip method was iterating over v4 and v6
subnets. So if it happened to pick a v6 subnet, it would cause the test to
request a floating IP address with a v6 IP address, which isn't valid.

Fix is up here: https://review.openstack.org/264599

On Wed, Jan 6, 2016 at 10:03 PM, Kevin Benton <blak111 at gmail.com> wrote:

> I spoke too soon. The test is racey and needs to be fixed, but that wasn't
> the cause of this failure. Floating IPs are being leaked as Clark
> suggested.  I'm looking into it now.
>
> On Wed, Jan 6, 2016 at 9:09 PM, Kevin Benton <blak111 at gmail.com> wrote:
>
>> Ugh, this test is racey. I should have paid closer attention before I
>> +2'ed, my bad.
>>
>> It just iterates external ports and uses an IP not in that list so if
>> another test grabbed the same IP, the floating IP creation request
>> specifying that IP will fail.
>>
>> Revert proposed here: https://review.openstack.org/#/c/264573/
>>
>> On Wed, Jan 6, 2016 at 9:00 PM, Clark Boylan <cboylan at sapwetik.org>
>> wrote:
>>
>>> On Wed, Jan 6, 2016, at 08:48 PM, Henry Gessau wrote:
>>> > Armando M. <armamig at gmail.com> wrote:
>>> > > Hi folks,
>>> > >
>>> > > Due to [1], Neutron related jobs (api, and lbaas) are failing.
>>> Please hold
>>> > > your +A button until the issue is resolved.
>>> > >
>>> > > Thanks,
>>> > > Armando
>>> > >
>>> > > [1] https://review.openstack.org/#/c/256164/
>>> >
>>> > That fix has merged, but now we have a new issue [2] in the api job. :(
>>> >
>>> > [2] https://bugs.launchpad.net/neutron/+bug/1531706
>>> >
>>> It appears that that test class at least is allocating unused floating
>>> IPs in resource_setup(). At the very least you can probably stop
>>> overallocating to give yourself some breathing room. I do not think this
>>> explains how an entire /24 of IPs gets used up so there may be a leak
>>> elsewhere.
>>>
>>> Clark
>>>
>>>
>>> __________________________________________________________________________
>>> OpenStack Development Mailing List (not for usage questions)
>>> Unsubscribe:
>>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>
>>
>>
>>
>> --
>> Kevin Benton
>>
>
>
>
> --
> Kevin Benton
>



-- 
Kevin Benton
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160106/12fcd8da/attachment.html>


More information about the OpenStack-dev mailing list