Open Stack

On Tue, Sep 13, 2016 at 12:53:41PM +0100, Derek Higgins wrote:
> On 9 September 2016 at 16:38, Paul Belanger <pabelanger at redhat.com> wrote:
> > Greetings,
> >
> > I would like to start the discussions around the removal of the bastion host
> > that sits in front of tripleo-test-cloud-rh1.  It is my understanding, all
> > traffic from tripleo-test-cloud-rh1 flows through this linux box.  Obviously
> > this is problematic for a public cloud.
> >
> > I currently do not know the history of the bastion host, I am hoping this thread
> > will start discussions around it.
> >
> > However, my personal preference is to remove the bastion from the pipeline
> > between internet and tripleo-test-cloud-rh1. My main objection to the host, is
> > the fact we do packet filtering of traffic flowing between the internet and
> > tripleo-test-cloud-rh1.
> 
> Would it be enough to simply remove the traffic filtering? or are
> there other problems you are hoping to get rid of?
> 
I'm hoping to remove it so we don't have to worry about managing it.  Today, it
exists for a reason (which I am still trying to understand). As I understand
today, it is used as a jump box to access some resource in
tripleo-test-cloud-rh1, assuming the controller and compute nodes.  But is that
all?

When things are working, there usually isn't problems. However, when problems
occur, it is another box that we have to loop into to trace outbound
connections. I think the most recent issues with DNS were a result of something
throttling connections?

> >
> > Ideally tripleo-test-cloud-rh1 will simply have an unfiltered network drop on
> > the public web, this is how we do it today with the infracloud in
> > #openstack-infra.
> >
> > This will avoid the need to gain access to a private server (bastion) and need
> > to manipulate networking traffic.
> >
> > I'd like for us to try and establish a time frame to make this happen too.
> 
> I don't know how much work this would be and what problems we would
> hit, historically the upstream tripleo team have been hands off when
> it comes to this box(and the rack switch), from our point of view we
> use it as a jump host to get to the other hosts on which openstack
> runs. And all outside traffic goes through it, I suppose the
> alternative would be to route the traffic directly to the overcloud
> controller.
> 
Yes, I think that would be the ideal setup. Removing any unmanaged linux systems
between tripleo-test-cloud-rh1 and the internet is the goal. This give tripleo
move control of the network and 1 less thing to depend on.

> We should be moving all our cloud usage onto RDO-Cloud some day, we
> should probably try and first get a timeline for when we are moving
> onto RDO-Cloud, if that is coming up soon perhaps we can just wait at
> this situation goes away.
> 
Pushing it off is one option, however I am offering up bandwidth to support this
effort.  Last I have heard RDO Cloud is still 3-6 months away.

> >
> > ---
> > Paul
> >
> > __________________________________________________________________________
> > 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
> 
> __________________________________________________________________________
> 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