Hi Dave,

In my case we have a kickstart server with some custom script handling preparing compute nodes. It kickstarted the machine and set up the interface/bridges/ips. My ip address design is based on a computer number like compute-10 (10.30.0.X).

Once my compute is ready then i go ahead and run playbook to add them in openstack. We do add machines that are bulk like 100 computers in one shot and it works out pretty well.

On Thu, Jun 30, 2022 at 1:22 PM Dmitriy Rabotyagov noonedeadpunk@gmail.com wrote:

Hi Dave,

No, unfortunately we don't have anything like that. Basically the reason is that OSA is super flexible and there's really a lot of ways on how to configure networking, depending on workload.

Not saying that some ppl would leverage bare metal provisioning systems, like Maas or Ironic, which would configure networking their own way.

However we have a role that is super helpful if you want to configure networking with systemd-networkd. We use that role in CI and have some example on it's usage here:

https://opendev.org/openstack/openstack-ansible/src/branch/master/tests/role...

So depending on your workload you just need to have a sample playbook and vars defined for hosts

чт, 30 июн. 2022 г., 18:42 Dave Hall kdhall@binghamton.edu:

...

Hello,

I am wondering if there is any sort of playbook or script for setting up the networking on a new physical host prior to adding it to a cluster. It is a bit cumbersome to do it by hand, especially given that bridges on one type of host need an IP while the same bridge on another type of host should not have an IP. Seeding the /etc/hosts file would be nice as well.

Going a bit further, it would be awfully nice if a given host could have the same 4th octet for all host interfaces, or at least for the important ones. I believe that since the 172.29 subnets are /22 this should be possible at least for smaller clusters.

Thanks.

-Dave

-- Dave Hall Binghamton University kdhall@binghamton.edu