Open Stack

Hi Fabian,


We are running Pike.

We have DVR with centralized SNAT on L3-HA routers. We also have L2
population enabled which I think is responsible for the issue.

We can reproduce the issue almost every time we are spawning 100 VMs at
once. Usually 3-6 VMs have no connectivity to services running on
network nodes like DHCP, DNS so they don't have network configured.


As I mentioned in the bug we also observing issues with L3-HA routers
provisioned using Heat. After debugging I'm almost sure there is a race
condition which is triggered by massive network resources provisioning.
When we provision routers manually we always have a working router.



On 19.03.2018 15:37, Fabian Zimmermann wrote:
> Hi Piotr,
>
> yes, we still debugging this.
>
> Im currently trying to reproduce the issue in our dev-env, but without
> luck so far.
>
> Are you able to reproduce the issue?
>
> We are running on ocata and currently trying to upgrade to pike
> followed by queens as fast as possible.
>
> What kind of neutron implementation do you use? (f.e. DVR with
> network-nodes for SNAT)?
>
> IRC: #openstack -> devfaz
>
> Thanks a lot,
>
>  Fabian
>
> Am 18.03.2018 um 22:37 schrieb Piotr Misiak:
>> Hi Fabian,
>>
>>
>> We also encounter issues with OVS table 22.
>>
>> Please take a look at this bug:
>> https://bugs.launchpad.net/neutron/+bug/1754695
>>
>> We are still debugging this.
>>
>> Maybe you hit the same bug?
>>
>>
>>
>> On 07.03.2018 08:56, Fabian Zimmermann wrote:
>>> Hi,
>>>
>>> sorry, it is table 22.
>>>
>>>   Fabian
>>>
>>>
>>> _______________________________________________
>>> OpenStack-operators mailing list
>>> OpenStack-operators at lists.openstack.org
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>
>> _______________________________________________
>> OpenStack-operators mailing list
>> OpenStack-operators at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>