7 * ``cluster``: a set of monitors
8 * ``quorum``: an active set of monitors consisting of a majority of the cluster
10 In order to initialize a new monitor, it must always be fed:
14 #. a cluster fsid (uuid)
16 In addition, a monitor needs to know two things:
18 #. what address to bind to
19 #. who its peers are (if any)
21 There are a range of ways to do both.
26 The logical id should be unique across the cluster. It will be
27 appended to ``mon.`` to logically describe the monitor in the Ceph
28 cluster. For example, if the logical id is ``foo``, the monitor's
29 name will be ``mon.foo``.
31 For most users, there is no more than one monitor per host, which
32 makes the short hostname logical choice.
37 The ``mon.`` secret key is stored a ``keyring`` file in the ``mon data`` directory. It can be generated
40 ceph-authtool --create-keyring /path/to/keyring --gen-key -n mon.
42 When creating a new monitor cluster, the keyring should also contain a ``client.admin`` key that can be used
43 to administer the system::
45 ceph-authtool /path/to/keyring --gen-key -n client.admin --set-uid=0 --cap mon 'allow *' --cap osd 'allow *' --cap mds 'allow'
47 The resulting keyring is fed to ``ceph-mon --mkfs`` with the ``--keyring <keyring>`` command-line argument.
52 The cluster fsid is a normal uuid, like that generated by the ``uuidgen`` command. It
53 can be provided to the monitor in two ways:
55 #. via the ``--fsid <uuid>`` command-line argument (or config file option)
56 #. via a monmap provided to the new monitor via the ``--monmap <path>`` command-line argument.
61 The monitor address can be provided in several ways.
63 #. via the ``--public-addr <ip[:port]>`` command-line option (or config file option)
64 #. via the ``--public-network <cidr>`` command-line option (or config file option)
65 #. via the monmap provided via ``--monmap <path>``, if it includes a monitor with our name
66 #. via the bootstrap monmap (provided via ``--inject-monmap <path>`` or generated from ``--mon-host <list>``) if it includes a monitor with no name (``noname-<something>``) and an address configured on the local host.
71 The monitor peers are provided in several ways:
73 #. via the initial monmap, provided via ``--monmap <filename>``
74 #. via the bootstrap monmap generated from ``--mon-host <list>``
75 #. via the bootstrap monmap generated from ``[mon.*]`` sections with ``mon addr`` in the config file
76 #. dynamically via the admin socket
78 However, these methods are not completely interchangeable because of
79 the complexity of creating a new monitor cluster without danger of
85 There are three basic approaches to creating a cluster:
87 #. Create a new cluster by specifying the monitor names and addresses ahead of time.
88 #. Create a new cluster by specifying the monitor names ahead of time, and dynamically setting the addresses as ``ceph-mon`` daemons configure themselves.
89 #. Create a new cluster by specifying the monitor addresses ahead of time.
95 Generate a monmap using ``monmaptool`` with the names and addresses of the initial
96 monitors. The generated monmap will also include a cluster fsid. Feed that monmap
97 to each monitor daemon::
99 ceph-mon --mkfs -i <name> --monmap <initial_monmap> --keyring <initial_keyring>
101 When the daemons start, they will know exactly who they and their peers are.
107 The initial monitor addresses can be specified with the ``mon host`` configuration value,
108 either via a config file or the command-line argument. This method has the advantage that
109 a single global config file for the cluster can have a line like::
111 mon host = a.foo.com, b.foo.com, c.foo.com
113 and will also serve to inform any ceph clients or daemons who the monitors are.
115 The ``ceph-mon`` daemons will need to be fed the initial keyring and cluster fsid to
116 initialize themselves:
118 ceph-mon --mkfs -i <name> --fsid <uuid> --keyring <initial_keyring>
120 When the daemons first start up, they will share their names with each other and form a
126 In dynamic "cloud" environments, the cluster creator may not (yet)
127 know what the addresses of the monitors are going to be. Instead,
128 they may want machines to configure and start themselves in parallel
129 and, as they come up, form a new cluster on their own. The problem is
130 that the monitor cluster relies on strict majorities to keep itself
131 consistent, and in order to "create" a new cluster, it needs to know
132 what the *initial* set of monitors will be.
134 This can be done with the ``mon initial members`` config option, which
135 should list the ids of the initial monitors that are allowed to create
138 mon initial members = foo, bar, baz
140 The monitors can then be initialized by providing the other pieces of
141 information (they keyring, cluster fsid, and a way of determining
142 their own address). For example::
144 ceph-mon --mkfs -i <name> --mon-initial-hosts 'foo,bar,baz' --keyring <initial_keyring> --public-addr <ip>
146 When these daemons are started, they will know their own address, but
147 not their peers. They can learn those addresses via the admin socket::
149 ceph daemon mon.<id> add_bootstrap_peer_hint <peer ip>
151 Once they learn enough of their peers from the initial member set,
152 they will be able to create the cluster.
158 Cluster expansion is slightly less demanding than creation, because
159 the creation of the initial quorum is not an issue and there is no
160 worry about creating separately independent clusters.
162 New nodes can be forced to join an existing cluster in two ways:
164 #. by providing no initial monitor peers addresses, and feeding them dynamically.
165 #. by specifying the ``mon initial members`` config option to prevent the new nodes from forming a new, independent cluster, and feeding some existing monitors via any available method.
167 Initially peerless expansion
168 ----------------------------
170 Create a new monitor and give it no peer addresses other than it's own. For
173 ceph-mon --mkfs -i <myid> --fsid <fsid> --keyring <mon secret key> --public-addr <ip>
175 Once the daemon starts, you can give it one or more peer addresses to join with::
177 ceph daemon mon.<id> add_bootstrap_peer_hint <peer ip>
179 This monitor will never participate in cluster creation; it can only join an existing
182 Expanding with initial members
183 ------------------------------
185 You can feed the new monitor some peer addresses initially and avoid badness by also
186 setting ``mon initial members``. For example::
188 ceph-mon --mkfs -i <myid> --fsid <fsid> --keyring <mon secret key> --public-addr <ip> --mon-host foo,bar,baz
190 When the daemon is started, ``mon initial members`` must be set via the command line or config file::
192 ceph-mon -i <myid> --mon-initial-members foo,bar,baz
194 to prevent any risk of split-brain.