[rhos-list] Setting Up Public/Private Networks Following Packstack Install

[Kashyap Chamarthy]

On Mon, Jul 14, 2014 at 09:44:03PM -0400, Perry Myers wrote:
> On 07/10/2014 04:45 PM, Lodgen, Brad wrote:
> > I completely rebuilt my environment (still using RHEL6.5 VM's on
> > vSphere 5.5 hosts) to use nova networking, hoping to make it a bit
> > simpler. This is likely to be my answer file being incorrect for my
> > environment, at least I'm hoping that since I'm going into the exam
> > tomorrow. Heh. Attached my sanitized answer file.
> > 
> > 
> > Start up a fresh instance with a RHEL6.5 KVM image and get these
> > errors during startup:
> > 
> > -Starting udev: udevd[377]: can not read '/etc/udev/rules.d/75-persistent-net-generator.rules'
> > Udevd[377]: can not read '/etc/udev/rules.d/75-persistent-net-generator.rules'
> > udevadm settle - timeout of 180 seconds, the event queue contains: lots of stuff I couldn't type fast enough
> > Udevadm settle - timeout of 0 seconds reached, the event queue contains:/sys/devices/virtual/net/lo (1317)
> > Udev still not settled. Waiting
> > -Did not get any network settings for eth0
> > -can't reach 169.254.169.254 for metadata [Network is unreachable]
> > -Getting data from cloudinit failed

So, these are resolved per your latest response.

For later reference (if you're using Neutron), I usually check if I have
relevant IP tables rules for metadaproxy, by checking on the router
namespace:

  ip netns exec qrouter-d72adddf-4c02-4916-ae6d-16bfdaf59d99 \
      iptables -L -t nat | grep 169
    REDIRECT   tcp  --  anywhere             169.254.169.254      tcp dpt:http redir ports 8700


FWIW, I previously used these notes to configure Neutron metadaproxy --
http://kashyapc.fedorapeople.org/virt/openstack/enabling-metadata-service.txt

> > When loading a cirrOS image, it just keeps rebooting over and over:
> > 
> > --can't find a bootable hard disk
> > --gpxe times out on DHCP
> > --no more network devices
> > --no bootable device, retrying in 60 seconds

(This too, I presume is resolved per your latest response to this thread.

> Ok, you mentioned in later thread that the Cirros issues were related to
> using the wrong image.
> 
> So, just to confirm... with the right Cirros image you're able to boot
> an instance on a Compute Node virtual machine running on an ESX host?
> 
> But when you boot a RHEL image on a Compute Node virtual machine running
> on an ESX host you get the above issues?

Yes, confirmation of this would be helpful. If you have more issues w/
RHEL image, we can try to diagnose it further w/ relevant log details.

> 
> Lars/Kashyap you guys have any thoughts?
> 
> Perry

-- 
/kashyap