[rdo-users] [rdo][ussuri][TripleO][nova][kvm] libvirt.libvirtError: internal error: process exited while connecting to monitor

Ruslanas Gžibovskis ruslanas at lpic.lt
Mon Jul 6 08:46:13 UTC 2020


Hi Alfredo,

since you mentioned, it is not essential to have that opstool, so I have
replaced it with
"sysstat"  /usr/share/tripleo-puppet-elements/overcloud-opstools/pkg-map so
now it is:
  "default": {
    "oschecks_package": "sysstat"
  }

And you are absolutely right regarding delorean, it took only OSP packages
from that, and kvm and libvirt are at your specified versions.

And then I believe, I found case for failing VM:

"6536f105-3f38-41bd-9ddd-6702d23c4ccb] Instance failed to spawn:
nova.exception.PortBindingFailed: Binding failed for port
af8ecd79-ddb8-4ba1-990d-1ccdb76f1442, please check"

so, my question is:
I have only control (pxe) network, which is distributed between sites and
OSP is having only one network (ControlPlane). How my controller and
compute network should look like?
My controller network looks like [1] and compute like [2]. When I uncomment
in compute br-provider part, it do not deploy.
does br-provider networks MUST be interconnectable?

I would need to have the possibility with the local network (vxlan) to
communicate between instances within the cloud, and external connectivity
would be done using provider vlan. each provider VLAN will be used only on
one compute node.  is it possible?


[0] http://paste.openstack.org/show/lUAOzDZdzCCcDrrPCASq/ # full package
list in libvirt container
[1] http://paste.openstack.org/show/795562/ # controller net-config
[2] http://paste.openstack.org/show/795563/ @ compute net-config

On Thu, 2 Jul 2020 at 17:36, Alfredo Moralejo Alonso <amoralej at redhat.com>
wrote:

>
>
> On Thu, Jul 2, 2020 at 4:38 PM Ruslanas Gžibovskis <ruslanas at lpic.lt>
> wrote:
>
>> it is, i have image build failing. i can modify yaml used to create
>> image. can you remind me which files it would be?
>>
>>
> Right, I see that the patch must not be working fine for centos and the
> package is being installed from delorean repos in the log.  I guess it
> needs an entry to cover the centos 8 case (i'm checking with opstools
> maintainer).
>
> As workaround I'd propose you to use the package from:
>
>
> https://trunk.rdoproject.org/centos8-ussuri/component/cloudops/current-tripleo/
>
> or alternatively applying some local patch to tripleo-puppet-elements.
>
>
>> and your question, "how it can impact kvm":
>>
>> in image most of the packages get deployed from deloren repos. I believe
>> part is from centos repos and part of whole packages in
>> overcloud-full.qcow2 are from deloren. so it might have bit different minor
>> version, that might be incompactible... at least it have happend for me
>> previously with train release so i used tested ci fully from the
>> beginning...
>> I might be for sure wrong.
>>
>
> Delorean repos contain only OpenStack packages, things like nova, etc...
> not kvm or things included in CentOS repos. KVM will always installed which
> should be installed from "Advanced Virtualization" repository. May you
> check what versions of qemu-kvm and libvirt you got installed into the
> overcloud-full image?, it should match with the versions in:
>
>
> http://mirror.centos.org/centos/8/virt/x86_64/advanced-virtualization/Packages/q/
>
> like qemu-kvm-4.2.0-19.el8.x86_64.rpm and libvirt-6.0.0-17.el8.x86_64.rpm
>
>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20200706/36a3c194/attachment-0001.html>


More information about the openstack-discuss mailing list