Open crawford opened 9 years ago
Comment by adamkdean Thursday Dec 04, 2014 at 16:38 GMT
One way around this is to create a dummy network, say 10.0.0.0 and put that in as eth1, and then change the etcd addr
and peer-addr
to $public_ipv4
.
We should be able to resolve this once we move to Ignition on Rackspace.
Any word on this? Would really like to be able to spin coreos machines up without a hitch and this appears to be the last hitch.
@ChrisMcKenzie Unfortunately, no. This has been a fundamental issue with CoreOS on Rackspace for over two years now. Once we add support for Rackspace to Ignition, this should no longer be an issue, but that is not a very high priority for us right now.
yeah currently what has to be done is boot the machine up login change the bash script to not look for eth1 and then restart the coreos-setup-environment
service and everything is happy. would a pull request be welcome for this script if I figure out a clean way to detect the interfaces?
@ChrisMcKenzie Absolutely! We'll take all the help we can get with CoreOS on Rackspace.
Issue by adamkdean Thursday Dec 04, 2014 at 14:43 GMT Originally opened as https://github.com/coreos/coreos-overlay/issues/981
On Rackspace's public cloud, if I create a CoreOS instance without PublicNet, CoreOS gets stuck in a loop waiting for both eth0 and eth1 (https://github.com/coreos/coreos-overlay/blob/master/coreos-base/oem-rackspace/files/coreos-setup-environment).
Not every CoreOS instance wants to be sat on the internet, and I'd rather be able to sit behind a firewall that I route my traffic through. This is not easy with problems like this.