From 69adfedbefc30341c47f930a2a07a0dedb9aef86 Mon Sep 17 00:00:00 2001
From: Kamil Sambor <kamil.sambor@gmail.com>
Date: Fri, 28 Jun 2024 11:25:37 +0200
Subject: [PATCH] Add docs how to run local nova code

---
 doc/testing_with_nova.md | 117 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 117 insertions(+)
 create mode 100644 doc/testing_with_nova.md

diff --git a/doc/testing_with_nova.md b/doc/testing_with_nova.md
new file mode 100644
index 000000000..1d1fd741d
--- /dev/null
+++ b/doc/testing_with_nova.md
@@ -0,0 +1,117 @@
+
+# Testing with local nova
+
+
+Nova is included in several services in the control plane: nova-api, nova-cellX-conductor, nova-cellX-novncproxy, nova-metadata, and nova-scheduler container image. The nova-operator doesn't support extra volume mounts for any of the mentioned pods. However, thanks to oc debug, it is possible to develop and test local changes to nova without having to build and deploy new Nova OpenStack changes.
+
+## Provide NFS access to your nova-openstack directory
+
+
+The technique described here uses NFS to access the nova opensctack directory on
+your development system, so you'll need to install an NFS server and create
+an appropriate export on your development system. Of course, this implies
+your OpenShift deployment that runs the openstack-operator has access to
+the NFS server, including any required firewall rules.
+
+When using OpenShift Local (aka CRC), your export will be something like this:
+
+```
+    % echo "${HOME}/nova/nova 192.168.130.0/24(rw,sync,no_root_squash)" > /etc/exports
+
+    % exportfs -r
+```
+
+Make sure nfs-server and firewalld are started:
+
+
+```    % systemctl start firewalld
+    % systemctl start nfs-server
+```
+### tip::
+
+   CRC installs its own firewall rules, which likely will need to be adjusted
+   depending on the location of your NFS server. If your nova-openstack
+   directory is on the same system that hosts your CRC, then the simplest
+   thing to do is insert a rule that essentially circumvents the other rules:
+
+   `% nft add rule inet firewalld filter_IN_libvirt_pre accept`
+
+## Create nova openstack PV and PVC
+
+
+Create an NFS PV, and a PVC that can be mounted on the ansibleee pods.
+
+### note::
+
+   While it's possible to add an NFS volume directly to a pod, the default k8s
+   Security Context Constraint (SCC) for non-privileged pods does not permit
+   NFS volume mounts. The approach of using an NFS PV and PVC works just as
+   well, and avoids the need to fiddle with SCC policies.
+
+```
+    NFS_SHARE=<Path to your nova directory>
+    NFS_SERVER=<IP of your NFS server>
+    cat <<EOF >nova-openstack-storage.yaml
+    apiVersion: v1
+    kind: PersistentVolume
+    metadata:
+      # Do not use just "nova" for the metadata name!
+      name: nova-openstack-dev
+    spec:
+      capacity:
+        storage: 1Gi
+      volumeMode: Filesystem
+      accessModes:
+        - ReadOnlyMany
+      # IMPORTANT! The persistentVolumeReclaimPolicy must be "Retain" or else
+      # your code will be deleted when the volume is reclaimed!
+      persistentVolumeReclaimPolicy: Retain
+      storageClassName: nova-openstack
+      mountOptions:
+        - nfsvers=4.1
+      nfs:
+        path: ${NFS_SHARE}
+        server: ${NFS_SERVER}
+    ---
+    apiVersion: v1
+    kind: PersistentVolumeClaim
+    metadata:
+      name: nova-openstack-dev
+    spec:
+      storageClassName: nova-openstack
+      accessModes:
+        - ReadOnlyMany
+      resources:
+        requests:
+          storage: 1Gi
+    EOF
+
+    oc apply -f nova-openstacke-storage.yaml
+```
+
+## Add additional mounts to your  CR
+
+
+First, we need to get the correct pod definition to debug. From oc get pod, find the pod in which you want to replace Nova and then run:
+```
+oc debug --keep-labels=true pod/nova-<pod_name> -o yaml
+```
+
+Next, edit the file to add the Nova OpenStack PVC to the pod definition:
+
+```
+spec:
+  containers:
+    volumeMounts:
+      - mountPath: /usr/lib/python3.9/site-packages/nova
+        name: nova-openstack
+      name: config-data
+        volumes:
+        - name: nova-openstack
+          persistentVolumeClaim:
+            claimName: nova-openstack
+            readOnly: true
+```
+
+Now we can spawn the container using: `oc debug -f <file_with_pod_definition>`
+Finally, scale your pod to zero by using `oc edit OpenStackControlPlane` and editing the replicas of the selected service to 0.