[openstack-dev] [Neutron] L3-HA - Solution for GW not pingable issue bug/1365461

Lubosz Kosnik lubosz.kosnik at intel.com
Fri Jan 22 09:05:55 UTC 2016


 From the performance perspective it will require additional CPU time 
because of the need to run that bash script.
I will try to prepare it as simple as possible but probability that it 
will kill server when there will be spawned huge amount of routers is 
not so big - after the implementation we can test CPU utilization with 
this validation enabled.

Cheers,
Lubosz (diltram) Kosnik

On 01/22/2016 09:43 AM, Anna Kamyshnikova wrote:
> Hi!
>
> This is great that you look into this issue! I think that first 
> solution looks more solid, my concern here is how does this will 
> affect the performance?
>
> On Thu, Jan 21, 2016 at 12:12 PM, Lubosz Kosnik 
> <lubosz.kosnik at intel.com <mailto:lubosz.kosnik at intel.com>> wrote:
>
>     He neutrinos,
>
>     Currently I'm working on this bug [1]. Almost one year ago Yoni
>     Shafrir prepared a patch to fix this issue but he got in review
>     information that this solution must be changed because it's using
>     only one script to check the GW availability and because of that
>     it cannot be used in multi tenat environment.
>     I tooked his code and was trying to upgrade that code to support
>     multiple scripts but I was designed separate solution for that.
>     I would like to know what do you think about this solution.
>
>     1. Add bash script generator to neutron/agent/linux/keepalived.py
>     2. There will be one script per one keepalived instance per node
>     3. There are two possible solutions for checking is everything is
>     working OK. Script will verify:
>         a. That all interfaces are up - internal router interfaces in
>     namespace, external interface taken from neutron configuration
>     file and also br-tun/br-int interfaces.
>         b. That GW is pingable from router NS - there is only one
>     problem what if GW is not configured in router already - plus we
>     could ping other network node or other server which IP is
>     specified in some configuration.
>
>     That solution will also fix this issue [2].
>     I would hear from you what do you think about that two possible
>     solutions and what do you think about whole solution at all.
>
>     Cheers,
>     Lubosz (diltram) Kosnik
>
>     [1] https://bugs.launchpad.net/neutron/+bug/1365461
>     [2] https://bugs.launchpad.net/neutron/+bug/1375625
>
>     __________________________________________________________________________
>     OpenStack Development Mailing List (not for usage questions)
>     Unsubscribe:
>     OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>     <http://OpenStack-dev-request@lists.openstack.org?subject:unsubscribe>
>     http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
>
>
> -- 
> Regards,
> Ann Kamyshnikova
> Mirantis, Inc
>
>
> __________________________________________________________________________
> 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

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160122/2446cdc3/attachment.html>


More information about the OpenStack-dev mailing list