Ask Your Question

bnemec's profile - activity

2019-05-06 11:10:47 -0500 answered a question error eventlet 0.24.1

It looks like os-win still caps eventlet. It should really stop doing that, but in the meantime you could try installing os-win manually before running devstack. As long as the version of eventlet it pulls in satisfies the other OpenStack projects' requirements that should get you around this.

2018-07-25 15:19:33 -0500 received badge  Supporter (source)
2018-07-25 14:07:22 -0500 received badge  Teacher (source)
2018-07-24 09:46:57 -0500 answered a question How to associate floating IP to specific interface with Heat

As I noted in my comment on Bernd's answer, I've been doing this for years and have never had the nics out of order, but it sounds like that's not something you can necessarily rely on.

I will also mention that in my experience putting two nics on the same network can create some weird routing issues in the instance, so if you can you may want to move the ovs port to a separate network, and then you'd be able to tell which nic is which by looking at the DHCP-assigned address.

Also be aware that most cloud images will only DHCP the first interface on the instance (as you can see in your ifconfig output), so you'll need to have your user_data script DHCP any extra interfaces.

2018-07-24 09:40:18 -0500 commented answer How to associate floating IP to specific interface with Heat

Interesting. I wasn't aware that the device ordering is not guaranteed. I will note that in several years of creating many Heat stacks on libvirt/kvm-based clouds I've never had the nics not come up in the order I specified. Apparently YMMV on that though.

2018-04-30 09:34:41 -0500 answered a question Why does openstack Nova use oslo_concurrency lockutils?

Nova does not use coroutines. It uses greenthreads from the eventlet library for concurrency within a process. They yield implicitly on IO rather than explicitly like a coroutine would (at least as I understand the difference). It does use multiple worker processes to get around the limitations of the Python GIL, but since this lock doesn't specify external=True (which would lock across processes instead of just threads) I don't think the multiple processes are relevant here.

2017-01-18 11:10:01 -0500 answered a question Tripleo in virtual environment

Virtual environment deployments of TripleO are not at all production ready, at least not with the documented instack-virt-setup method. While it may be possible to do a production deployment using TripleO and some nodes deployed to VMs, it actually ends up being more complicated if you want real HA, and it introduces considerably more potential points of failure. In addition, deploying a compute node in a VM introduces nested virt problems.

Deploying to virtual environments can be a good way to learn the basics of how TripleO works, but any deployments you intend to use for productive work should be done to baremetal.

2015-03-09 13:41:58 -0500 answered a question instack-install-undercloud fails with glance: error: unrecognized arguments: --public

Are you sure it's instack-install-undercloud that's failing? I believe that call actually happens in instack-prepare-for-overcloud: https://github.com/rdo-management/instack-undercloud/blob/stable/juno/scripts/instack-prepare-for-overcloud#L48 (https://github.com/rdo-management/ins...)

The workaround is to edit /usr/bin/instack-prepare-for-overcloud and change "--public" to "--is-public True". Note that there are actually 4 instances of --public in that file, and they will all need to be changed.