On Fri, Feb 1, 2019 at 10:51 PM Dominik Holler <dholler@redhat.com> wrote:

On Fri, 1 Feb 2019 22:35:00 +0200
Dan Kenigsberg <danken@redhat.com> wrote:

> On Fri, Feb 1, 2019 at 10:18 PM Dominik Holler <dholler@redhat.com>
> wrote:
> >
> > On Fri, 1 Feb 2019 14:37:10 +0100
> > Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
> >
> > > Hello,
> > > at this moment (about two days ago) I have updated only engine
> > > (external, not self hosted) from 4.2.7.5 to 4.2.8.2
> > >
> > > As soon as I'm starting for the first time a VM with an ovn based
> > > nic I get what below in ovirt-provider-ovn.log
> > >
> > > In admin gui, if I try for example to start via "run once" I get:
> > > "
> > > Error while executing action Run VM once: Failed to communicate
> > > with the external provider, see log for additional details.
> > > "
> > > Any clue?
> >
> > The ovirt-provider-ovn fails during checking the credentials at
> > engine's sso because of a networking problem.
>
> That would be odd - after all we're using the loopback interface
> From: ::ffff:127.0.0.1:49582 Request: GET /v2.0/ports
> but please try the url.
>

Communication from Engine to ovirt-provider-ovn via OpenStack API looks
good. The problem seems to be in the communication from
ovirt-provider-ovn to engine's sso.
The hostname to resolve seems to be 'engine-host'.

Hi all, so it seems that updating from 4.2.7 to 4.2.8 wiped the file entirely...

In fact now I only have:

 [root@ovmgr1 ~]# ll /etc/ovirt-provider-ovn/conf.d/

total 4

-rw-r--r--. 1 root root 194 Jan 14 16:18 README

[root@ovmgr1 ~]# 

While on another similar environment in 4.2.7 I still have:

[root@ovirt ~]# ll /etc/ovirt-provider-ovn/conf.d/

total 8

-rw-r--r--. 1 root root 561 Dec 31  2017 10-setup-ovirt-provider-ovn.conf

-rw-r--r--. 1 root root 194 Oct 17 05:56 README

[root@ovirt ~]#

So in my updated environment I think it tries to use the default file values in /etc/ovirt-provider-ovn/ovirt-provider-ovn.conf

[OVIRT]

ovirt-host=https://engine-host

The question now is why it has been wiped out?

This environment has external VM working as ovirt engine, so not SHE.

I see that engine-backup doesn't run backup of the /etc/ovirt-provider-ovn/ ... could it be useful to include it?

I should have a VSPhere image backup of the VM so I should be able to restore the file. I'll see...

In the mean time could it be the update of the package has been the responsible?

Eg, in my 4.2.7 env I currently have ovirt-provider-ovn-1.2.16-1.el7.noarch

The update to 4.2.8 in the other system has put in ovirt-provider-ovn-1.2.18-1.el7.noarch

HIH debugging

Gianluca