Ceph does not allow for more than one public network. If you have multiple
subnets that need to access ceph data, then you should be using a router
and firewall rules to route the traffic between them.

On Fri, Jul 21, 2017, 12:48 PM Yang X <[email protected]> wrote:

> Our existing setup is as follows and we won't be able to change the network
> configuration due to security limitations:
> client 1:   rbd devices on 153.64.X.X network (1GE network)
> client 2:   rbd devices on 10.25.X.X network (10GE fast switch)
> single monitor and MDS server multihomed on both 153.64.X.X and
> 10.25.X.X network.
> OSDs are also both on the 10GE fast network 10.25.X.X and slow
> 153.64.X.X network.
>
> Our problem is that when I tried to specify multiple public networks
> separated by comma in the ceph.conf file, the OSD nodes would only
> communicate through the first network tuple specified.
>
> [global]
> mon host = 153.64.109.25
> cluster network = 10.25.0.0/16
> public network = 10.25.0.0/16,153.64.0.0/16
>
> The idea is to allow rbd block devices on client 2 to perform faster
> via 10GE network and at the same time to allow rbd devices on client1
> to work via another network.
>
> However,  client 2 would not connect with the above monitor for
> obvious reason. and client 1 would connect but "rbd" commands would
> fail because it is trying to communicate to OSDs with 10.25.0.0
> network where it is not accessible from client1.
>
> Anybody has any experience to  have made the "comma separated"
> multiple public network work.
>
> And also I wonder if it is possible to have the monitor daemon to
> listen on multiple interfaces.
> such as:
> mon host = 153.64.109.25,10.25.3.8
>
> Your help in this will be much appreciated!!
>
> Sincerely,
>
> Yang
> _______________________________________________
> ceph-users mailing list
> [email protected]
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to