Merge pull request #3842 from mboersma/minimum-three

docs(installing_deis): etcd doesn't require an odd # of nodes

Author: mboersma

docs/installing_deis/aws.rst

@@ -70,10 +70,7 @@ By default, the script will provision 3 servers. You can override this by settin
 
     $ export DEIS_NUM_INSTANCES=5
 
-Note that for scheduling to work properly, clusters must consist of at least 3 nodes and always
-have an odd number of members. For more information, see `etcd disaster recovery`_.
-
-Deis clusters of less than 3 nodes are unsupported.
+A Deis cluster must have 3 or more nodes. See :ref:`cluster-size` for more details.
 
 
 Generate a New Discovery URL

docs/installing_deis/azure.rst

@@ -76,12 +76,11 @@ With that, let's run the azure-coreos-cluster script which will create the CoreO
          --data-disk
          --custom-data azure-user-data
 
-This script will by default provision a 3 node cluster but you can increase this with the
+This script will by default provision a 3-node cluster, but you can increase this with the
 ``--num-nodes`` parameter. Likewise, you can increase the VM size using ``--vm-size``.
 It is not recommended that you use smaller than Large (A3) sized instances.
 
-Note that for scheduling to work properly, clusters must consist of at least 3 nodes and always
-have an odd number of members. For more information, see `etcd disaster recovery`_.
+A Deis cluster must have 3 or more nodes. See :ref:`cluster-size` for more details.
 
 
 Configure DNS

docs/installing_deis/openstack.rst

@@ -95,9 +95,7 @@ Generate a New Discovery URL
 Choose number of instances
 --------------------------
 
-For scheduling to work properly, clusters must consist of at least 3 nodes and always have an odd
-number of members. Please refer to :ref:`system-requirements` for more information about cluster
-size requirements.
+A Deis cluster must have 3 or more nodes. See :ref:`cluster-size` for more details.
 
 Instruct the provision script to launch the desired number of nodes:
 

docs/installing_deis/rackspace.rst

@@ -73,10 +73,8 @@ By default, the provision script will provision 3 servers. You can override this
 
     $ DEIS_NUM_INSTANCES=5 ./provision-rackspace-cluster.sh deis-key
 
-Note that for scheduling to work properly, clusters must consist of at least 3 nodes and always have an odd number of members.
-For more information, see `etcd disaster recovery`_.
+A Deis cluster must have 3 or more nodes. See :ref:`cluster-size` for more details.
 
-Deis clusters of less than 3 nodes are unsupported.
 
 Run the Provision Script
 ------------------------

docs/installing_deis/system-requirements.rst

@@ -28,13 +28,17 @@ applications.
 Running smaller machines will likely result in increased system load and has been known to result
 in component failures, issues with etcd/fleet, and other problems.
 
+.. _cluster-size:
+
 Cluster size
 ------------
 
-For scheduling to work properly, clusters must consist of at least three nodes and always
-have an odd number of members. This is mostly because the underlying CoreOS cluster must always
-be able to obtain a quorum (see `etcd disaster recovery`_). Additionally, the :ref:`Store`
-component keeps three replicas of stored data, so it requires at least three nodes.
+For :ref:`scheduling <choosing_a_scheduler>` and the :ref:`deis-store <Store>` components
+to work properly, clusters must have at least three nodes. The ``etcd`` service must always
+be able to obtain a quorum, and the Ceph data store must maintain at least three replicas
+of persistent data.
+
+See `optimal etcd cluster size`_ and `etcd disaster recovery`_ for further information.
 
 .. important::
 
@@ -50,5 +54,6 @@ Network
 .. include:: ../_includes/_private-network.rst
 
 .. _`dokku`: https://github.com/progrium/dokku
+.. _`optimal etcd cluster size`: https://coreos.com/docs/cluster-management/scaling/etcd-optimal-cluster-size/
 .. _`etcd disaster recovery`: https://github.com/coreos/etcd/blob/master/Documentation/admin_guide.md#disaster-recovery
 .. _`sponsored`: http://deis.io/deis-sponsors-dokku/