Clarify that you cannot abort an upgrade

docs/reference/upgrade.asciidoc

@@ -64,21 +64,6 @@ To upgrade directly to {version} from 6.6 or earlier, you must shut down the
 cluster, install {version}, and restart. For more information, see
 <<restart-upgrade, Full cluster restart upgrade>>.
 
-[float]
-== Preparing to upgrade
-
-Before upgrading {es}:
-
-. Check the <<deprecation-logging, deprecation log>> to see if you are using
-any deprecated features and update your code accordingly. By default,
-deprecation warnings are logged when the log level is set to `WARN`.
-. Review the <<breaking-changes,breaking changes>> and make any necessary changes
-to your code and configuration for {version}.
-. If you use custom plugins, make sure compatible versions are available.
-. Test upgrades in a dev environment before upgrading your production cluster.
-. <<modules-snapshots,Back up your data!>> You must have a snapshot of your
-data to roll back to an earlier version.
-
 --
 
 include::upgrade/rolling_upgrade.asciidoc[]

docs/reference/upgrade/cluster_restart.asciidoc

@@ -9,7 +9,12 @@ NOTE: If you are running a version prior to 6.0,
 and reindex your old indices or bring up a new {version} cluster and
 <<reindex-upgrade-remote, reindex from remote>>.
 
-To perform a  full cluster restart upgrade to {version}:
+include::preparing_to_upgrade.asciidoc[]
+
+[float]
+=== Upgrading your cluster
+
+To perform a full cluster restart upgrade to {version}:
 
 . *Disable shard allocation.*
 +
@@ -52,6 +57,11 @@ include::upgrade-node.asciidoc[]
 include::set-paths-tip.asciidoc[]
 --
 
+If upgrading from a 6.x cluster, you must also
+<<modules-discovery-bootstrap-cluster,configure cluster bootstrapping>> by
+setting the <<initial_master_nodes,`cluster.initial_master_nodes` setting>> on
+the master-eligible nodes.
+
 . *Upgrade any plugins.*
 +
 Use the `elasticsearch-plugin` script to install the upgraded version of each
@@ -70,10 +80,6 @@ If you have dedicated master nodes, start them first and wait for them to
 form a cluster and elect a master before proceeding with your data nodes.
 You can check progress by looking at the logs.
 
-If upgrading from a 6.x cluster, you must
-<<modules-discovery-bootstrap-cluster,configure cluster bootstrapping>> by
-setting the `cluster.initial_master_nodes` setting.
-
 As soon as enough master-eligible nodes have discovered each other, they form a
 cluster and elect a master. At that point, you can use
 <<cat-health,`_cat/health`>> and <<cat-nodes,`_cat/nodes`>> to monitor nodes

docs/reference/upgrade/preparing_to_upgrade.asciidoc

@@ -0,0 +1,28 @@
+[float]
+=== Preparing to upgrade
+
+It is important to prepare carefully before starting an upgrade. Once you have
+started to upgrade your cluster to version {version} you must complete the
+upgrade. As soon as the cluster contains nodes of version {version} it may make
+changes to its internal state that cannot be reverted. If you cannot complete
+the upgrade then you should discard the partially-upgraded cluster, deploy an
+empty cluster of the version before the upgrade, and restore its contents from
+a snapshot.
+
+Before you start to upgrade your cluster to version {version} you should do the
+following.
+
+. Check the <<deprecation-logging, deprecation log>> to see if you are using any
+deprecated features and update your code accordingly.
+
+. Review the <<breaking-changes,breaking changes>> and make any necessary
+changes to your code and configuration for version {version}.
+
+. If you use any plugins, make sure there is a version of each plugin that is
+compatible with {es} version {version}.
+
+. Test the upgrade in an isolated environment before upgrading your production
+cluster.
+
+. <<modules-snapshots,Back up your data by taking a snapshot!>>
+

docs/reference/upgrade/rolling_upgrade.asciidoc

@@ -23,7 +23,12 @@ Rolling upgrades are supported:
 Upgrading directly to {version} from 6.6 or earlier requires a
 <<restart-upgrade, full cluster restart>>.
 
-To perform a rolling upgrade from {prev-major-version} to {version}:
+include::preparing_to_upgrade.asciidoc[]
+
+[float]
+=== Upgrading your cluster
+
+To perform a rolling upgrade to {version}:
 
 . *Disable shard allocation*.
 +