We have an OpenShift cluster that is all internal, for dev purposes. It consists of a master + 2 nodes, all of which have a private IP. As long as the three hosts can communicate (are on the same network or can route to one another, AND can resolve each other’s hostnames in DNS) then you’re good. These three hosts run CentOS 7 Atomic. A wildcard DNS record points to the IPs of the two nodes, where the routers run, so its a round-robin DNS thing. Not the most optimal HA set up, but its simple and works well thus far.
Also, we used a regular CentOS 7 host (not Atomic) that’s not part of the cluster that acts as the “utility” server. It has Ansible installed and can ssh to all the cluster hosts in order to fire off the playbooks.
We are just getting started so haven’t set up our production cluster yet. For that, I imagine we’ll need public IPs (assigned to our nodes that are running routers) so that the world can get to our apps.
So anyway, I *think* the answer to your question is you need public IPs on the openshift nodes that run your routers if the applications are to be publicly accessible. The blog post here suggests running your router on just the master: https://tobrunet.ch/articles/dive-into-openshift-v3
…A wildcard DNS entry pointing to the IP address of the master. It is at this master where routers will run to allow for external clients to request application resources running within OpenShift.
Thanks for your reply. Just the main goal we want to obtain is to keep our traffic from pod to pod (using routes, router, dns-wildcard) internal. So performing al this stuf on a private IP. Is that
He is using public ip's + private ip's. Are the privates useful in this case?
We're able to use both and use and we can setup our own dns server but we don't want that our routes are going outside of our cluster. In public and than going back in the cluster.
So main goal: translations of routes through router should stay in the private network.
you may need internet connectivity. Public IPs is not a requirement for that (confer proxy and NAT). Another option is to install OpenShift disconnected. See: https://docs.openshift.com/container-platform/3.3/install_config/install/disconnected_install.html.
Also, editing etc/hosts is not enough. You will require a proper DNS server (dnsmasq for instance) as the containers don't use /etc/hosts of the host for name resolution.
On Thu, Dec 8, 2016 at 1:37 PM, Den Cowboy <dencowboy hotmail com> wrote:
Delivering value year after year
Information in this e-mail may be confidential. It is intended only for the addressee(s) identified above. If you are not the addressee(s), or an employee or agent of the addressee(s), please note that any dissemination, distribution, or copying of this communication is strictly prohibited. If you have received this e-mail in error, please notify the sender of the error.