From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nikolay Borisov Subject: Connectivity between containers with ipoib interface. Date: Tue, 9 Aug 2016 17:33:44 +0300 Message-ID: <57A9E9C8.5060905@kyup.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: "linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" List-Id: linux-rdma@vger.kernel.org Hello, I have multiple containers into which I have put "virtual" ipoib interface like so: ip link add link ib0 name ip1 type ipoib ip link set dev ip1 netns $(pid of container) The containers are setup with ip addresses from a different subnet than the "base" addresses on the host-side ib0 interface. With this setup containers can access other nodes on the infiniband network with no problems. However, if 2 containers on the same host node, with ipoib addresses from the same subnet try to access one another - this doesn't work. The only way to enable such communication is to play tricks with routing such that packets exit the container and host node, and are sent to another physical node on the infiniband network, which then routes them back to the initial host node. I'd like to avoid such gymnastics and have connectivity to containers on the same node without traffic going out the ib0 interface. I guess what I'm asking for is something similar to the bridge mode in macvlan/macvtap drivers. Regards, Nikolay -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html