You can run multiple Ceph clusters on the same hardware. They will have completely separate monitors, OSDs (including separate data disks and journals that will not be shared between clusters), MDSs etc. This provides a higher level of isolation than e.g. just using multiple RADOS pools and CRUSH rulesets. This multiple cluster support is activated in all the Ceph commands, such as ``ceph``, with the ``--cluster=`` option. For example, to see the status of the cluster ``uswest1a``, run ``ceph --cluster=uswest1a -s``. This will read ``/etc/ceph/uswest1a.conf`` instead of the default ``ceph.conf``, and use the monitor addresses and keyrings specified in that file. In the configuration file, the name of the cluster can be referred to as ``$cluster``. Cluster names must consist of letters a-z and digits 0-9 only. This also means that when preparing disks for OSD hotplugging, you may want to specify a non-default cluster with ``ceph-disk-prepare --cluster=NAME /dev/sdb``. To guard against typos in IP addresses and port numbers, make sure each ``CLUSTER.conf`` file specifies an ``fsid``. You should also have a distinct ``mon.`` key for each cluster. Following documented installation procedures makes sure they are both randomly generated for each cluster. Current status: multi-cluster support has not been fully QA'ed, and there are known issues (e.g. http://tracker.newdream.net/issues/3253 http://tracker.newdream.net/issues/3277 ). -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
