* Connectivity between containers with ipoib interface.
@ 2016-08-09 14:33 Nikolay Borisov
0 siblings, 0 replies; only message in thread
From: Nikolay Borisov @ 2016-08-09 14:33 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA
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
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2016-08-09 14:33 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-08-09 14:33 Connectivity between containers with ipoib interface Nikolay Borisov
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.