diff --git a/docs/operation/index.rst b/docs/operation/index.rst
index 3a51e3b5f8..a08c430b6e 100644
--- a/docs/operation/index.rst
+++ b/docs/operation/index.rst
@@ -20,6 +20,7 @@ do not have a working MetalK8s_ setup.
    solutions
    changing_node_hostname
    metalk8s-utils
+   registry_ha
    listening_processes
    troubleshooting/index
    sosreport
diff --git a/docs/operation/registry_ha.rst b/docs/operation/registry_ha.rst
new file mode 100644
index 0000000000..602e4a14b9
--- /dev/null
+++ b/docs/operation/registry_ha.rst
@@ -0,0 +1,78 @@
+Registry HA
+===========
+
+MetalK8s come with his own registry used to serve all images used by MetalK8s
+containers, so that it can run fully offline, this registry container sit on
+the Bootstrap node.
+
+Highly available registry allow to lose the Bootstrap node and still be able
+to schedule pods on nodes that need to pull some container images. So it
+means, lose of Bootstrap node should have no impact on the workload pods
+running in the cluster.
+
+.. note::
+
+  This procedure only talk about registry HA as Bootstrap HA is not
+  supported for the moment, so it's only a part of the Bootstrap
+  functionnaly. Check this ticket for more informations
+  https://github.com/scality/metalk8s/issues/2002
+
+Prepare the node
+----------------
+
+In order to be able to schedule a repository pod on a node that will
+be used as registry, you need this node to be part of the MetalK8s
+cluster, no specific roles or taints is needed, this pods can be scheduled
+on any node.
+
+All ISOs listed in ``archives`` section of
+``/etc/metalk8s/bootstrap.yaml`` and of ``/etc/metalk8s/solutions.yaml``
+from the Bootstrap node must be on the exact same place on the node
+that will be used as registry.
+
+Deploy the registry
+-------------------
+
+Connect on the node where you want to deploy this registry and run the
+following salt states
+
+- Prepare all the MetalK8s ISOs
+
+  .. parsed-literal::
+
+    root@node-1 $ salt-call state.sls \\
+        metalk8s.archives.mounted \\
+        saltenv=metalk8s-|version|
+
+- If you have some solutions, prepare solutions ISOs
+
+  .. parsed-literal::
+
+    root@node-1 $ salt-call state.sls \\
+        metalk8s.solutions.available \\
+        saltenv=metalk8s-|version|
+
+- Deploy the registry container
+
+  .. parsed-literal::
+
+    root@node-1 $ salt-call state.sls \\
+        metalk8s.repo.installed \\
+        saltenv=metalk8s-|version|
+
+
+Reconfigure all nodes
+---------------------
+
+In order to have registry HA we need to reconfigure all containerd to
+use both registry as endpoint so that if one is down containerd
+automatically pull from the other registry.
+
+For this you need to run a state on all node using Salt Master
+
+.. parsed-literal::
+
+  root@bootstrap $ kubectl exec -n kube-system -c salt-master \\
+     --kubeconfig=/etc/kubernetes/admin.conf \\
+     salt-master-bootstrap -- salt '*' state.sls \\
+     metalk8s.container-engine saltenv=metalk8s-|version|