Remove network infra VM

Since all this runs is the image cache now it can be removed, instead we'll use the podman container cache running on the hypervisor
alknopfler · Dec 13, 2023 · 4e0fb66 · 4e0fb66
1 parent b8c6ab8
commit 4e0fb66
Show file tree

Hide file tree

Showing 29 changed files with 74 additions and 449 deletions.
diff --git a/03_launch_mgmt_cluster.sh b/03_launch_mgmt_cluster.sh
@@ -0,0 +1,17 @@
+#!/usr/bin/env bash
+set -eux
+
+PROJECT_DIR=$(dirname -- $(readlink -e -- ${BASH_SOURCE[0]}))
+EXTRA_VARS_FILE=${EXTRA_VARS_FILE:-$PROJECT_DIR/extra_vars.yml}
+
+if [[ "$(id -u)" -eq 0 ]]; then
+  echo "Please run as a non-root user"
+  exit 1
+fi
+
+# Run ansible configure host playbook
+export ANSIBLE_ROLES_PATH=$PROJECT_DIR/roles
+ANSIBLE_FORCE_COLOR=true ansible-playbook \
+  -i ${PROJECT_DIR}/inventories/localhost_inventory.yml \
+  -e "@${EXTRA_VARS_FILE}" \
+  $PROJECT_DIR/playbooks/setup_metal3_core.yml $@
diff --git a/README.md b/README.md
@@ -1,7 +1,6 @@
 # Table Of Contents
 
 - [Overview](#overview)
-  - [Networking](#networking)
 - [Prerequisites](#prerequisites)
 - [How To Setup Metal3 Demo Environment](#how_to_setup_metal3_demo)
 
@@ -14,24 +13,12 @@ of [Kubernetes Cluster API][CAPI], for Kubernetes workload cluster
 life cycle management. The demo environment consist of two VMs,
 Metal3 Network Infra and Metal3 Core respectively.
 ![Metal3 Demo Overview](images/Metal3-Demo-Overview.png)
-As depicted by the diagram above, the Metal3 Network Infra VM is designed
-to emulate the infrastructure pieces, namely DNS, DHCP, and media server,
-which are required by Metal3 and typically expected to be deployed outside
-of the management cluster in a production environment. Metal3 Core VM has
+Metal3 Core VM has
 all the pieces, namely CAPI (Cluster API) controller, RKE2 bootstrap provider
 (CABPR), RKE2 control plane provider (CACPPR),
 Metal3 infrastructure provider (CAPM3), Baremetal Operator, and
 OpenStack Ironic, in a typical production Metal3 management cluster.
 
-## Networking <a name="networking" />
-
-For security purposes, network segmentation is expected in production
-environment, which usually consist of an internal provisioning network
-for bare metal provisioning, and public network which is routable to
-the internet. In the demo environment is used only one network and the host where
-the VMs are running is expected to have a networking bridge for the
-public network (i.e. tagged VLAN).
-
 # Prerequisites <a name="prerequisites" />
 
 * Host is expected to have one network bridge for the public network (i.e. tagged VLAN).
@@ -46,5 +33,4 @@ public network (i.e. tagged VLAN).
 - For the automation of this deployment, [click here.](./scripts/README.md)
 
 [CAPI]: https://cluster-api.sigs.k8s.io/introduction.html
-[cloud_init_network_config]: https://cloudinit.readthedocs.io/en/latest/reference/network-config.html
 [metal3]: https://github.com/metal3-io
diff --git a/common_setup.sh b/common_setup.sh
diff --git a/docs/example-manifests/dhcp/rke2-agent.yaml b/docs/example-manifests/dhcp/rke2-agent.yaml
@@ -58,10 +58,10 @@ spec:
         matchLabels:
           cluster-role: worker
       image:
-        checksum: http://media.suse.baremetal/openSUSE-Leap-15.5.x86_64-NoCloud.qcow2.md5
-        checksumType: md5
+        checksum: http://imagecache.local:8080/openSUSE-Leap-15.5.x86_64-NoCloud.metal3.qcow2.sha256
+        checksumType: sha256
         format: qcow2
-        url: http://media.suse.baremetal/openSUSE-Leap-15.5.x86_64-NoCloud.qcow2
+        url: http://imagecache.local:8080/openSUSE-Leap-15.5.x86_64-NoCloud.metal3.qcow2
 ---
 apiVersion: infrastructure.cluster.x-k8s.io/v1beta1
 kind: Metal3DataTemplate

diff --git a/docs/example-manifests/dhcp/rke2-control-plane.yaml b/docs/example-manifests/dhcp/rke2-control-plane.yaml
@@ -63,10 +63,10 @@ spec:
         matchLabels:
           cluster-role: control-plane
       image:
-        checksum: http://media.suse.baremetal/openSUSE-Leap-15.5.x86_64-NoCloud.qcow2.md5
-        checksumType: md5
+        checksum: http://imagecache.local:8080/openSUSE-Leap-15.5.x86_64-NoCloud.metal3.qcow2.sha256
+        checksumType: sha256
         format: qcow2
-        url: http://media.suse.baremetal/openSUSE-Leap-15.5.x86_64-NoCloud.qcow2
+        url: http://imagecache.local:8080/openSUSE-Leap-15.5.x86_64-NoCloud.metal3.qcow2
 ---
 apiVersion: infrastructure.cluster.x-k8s.io/v1beta1
 kind: Metal3DataTemplate

diff --git a/docs/setup/metal3-setup.md b/docs/setup/metal3-setup.md
@@ -46,26 +46,24 @@ If desired the defaults from `extra_vars.yml` can be customized, copy the file a
    ```
     - If you plan not to use the virsh networks, you will need to set up your own network bridges.
 
-3. Create the Network Infra VM
+3. Configure the host
 
-- In the main directory of the repository, execute the script to create the network-infra VM
+- In the main directory of the repository, execute the script to configure the host:
 
   ```shell
-  ./setup_metal3_network_infra.sh
+  ./02_configure_host.sh
   ```
 
-- You may pass `-vvv` at the end of the script to see the output of the script
-- The network-infra script must have completed without any errors before creating the core VM in step 8
-
-4. Create the core VM
+4. Create management cluster
 
   ```shell
-  ./setup_metal3_core.sh
+  ./03_launch_mgmt_cluster.sh
   ```
 
-- You may pass `-vvv` at the end of the script to see the output
-
-5. Assuming you are using the default configuration you can ssh into each of the VMs using the IPs below:
+5. Assuming you are using the default configuration you can ssh into the management cluster VM as follows:
 
 - Core VM Running Metal3: `ssh [email protected]` or `virsh console metal3-core`
-- Network Infra VM Running with public internet access: `ssh [email protected]` or `virsh console metal3-network-infra`
+
+## Development Notes
+
+- You may pass `-vvv` at the end of the scripts to see more verbose output, or to pass arbitrary additional arguments to ansible-playbook
diff --git a/docs/setup/rke2-cluster.md b/docs/setup/rke2-cluster.md
@@ -37,28 +37,12 @@ EOF
 virsh net-update egress add-last ip-dhcp-host host.xml --live
 ```
 
-4. Create an XML file containing the following
-
-```shell
-cat << EOF > ~/vbmc/dns.xml
-<host ip='192.168.125.100'>
-  <hostname>media.suse.baremetal</hostname>
-</host>
-EOF
-```
-
-5. Live update the egress network once again
-
-```shell
-virsh net-update egress add-last dns-host dns.xml --live
-```
-
-6. SSH into the metal3-core VM
+4. SSH into the metal3-core VM
 ```shell
 ssh [email protected]
 ```
 
-7. Download the example manifests
+5. Download the example manifests
 
 ```shell
 curl https://raw.githubusercontent.com/suse-edge/metal3-demo/main/docs/example-manifests/dhcp/rke2-control-plane.yaml > rke2-control-plane.yaml
@@ -69,13 +53,13 @@ curl https://raw.githubusercontent.com/suse-edge/metal3-demo/main/docs/example-m
   If you have made your own changes or have differences in your setup, you may need to update the manifests.
 - This configuration assumes DHCP-only network setup.
 
-8. Deploy the control plane
+6. Deploy the control plane
 
 ```shell
 kubectl apply -f rke2-control-plane.yaml
 ```
 
-9. Verify that the control plane is properly provisioned
+7. Verify that the control plane is properly provisioned
 
 ```shell
 $ clusterctl describe cluster sample-cluster
@@ -86,13 +70,13 @@ $ clusterctl describe cluster sample-cluster
   │ └─Machine/sample-cluster-chflc                        True                     23m
 ```
 
-10. Deploy the agent
+8. Deploy the agent
 
 ```shell
 kubectl apply -f rke2-agent.yaml
 ```
 
-11. Verify that the agent is properly provisioned and has successfully joined the cluster
+9. Verify that the agent is properly provisioned and has successfully joined the cluster
 
 ```shell
 $ clusterctl describe cluster sample-cluster

diff --git a/docs/setup/vbmh-setup.md b/docs/setup/vbmh-setup.md
@@ -85,7 +85,7 @@ EOF
 Add this line to /etc/hosts
 
 ```text
-192.168.125.100 media.suse.baremetal
+192.168.125.1 imagecache.local
 ```
 
 - This is necessary for DNS resolutions for Metal3 in the metal3-demo environment.
@@ -229,4 +229,4 @@ kubectl apply -f node2.yaml
 - Using `baremetal node list` may show `manageable` immediately after creating the nodes,
   but this is only temporary, we want to wait for it to say `manageable` after it has been inspected.
 - If there is an issue during the provisioning process. Take the UUID of the baremetal node
-  and run `baremetal node show UUID` for a detailed output on what might have gone wrong.  
+  and run `baremetal node show UUID` for a detailed output on what might have gone wrong.  
diff --git a/extra_vars.yml b/extra_vars.yml
@@ -23,7 +23,6 @@ rke2_channel_version: v1.24
 
 metal3_vm_libvirt_network_params: '--network bridge=m3-egress,model=virtio'
 
-metal3_network_infra_public_ip: 192.168.125.100
 vm_egress_gw: 192.168.125.1
 
 
@@ -32,18 +31,6 @@ enable_dhcp: true
 dhcp_router: 192.168.124.1
 dhcp_range: 192.168.124.150,192.168.124.180
 
-metal3_network_infra_vm_network:
-  version: 2
-  ethernets:
-    eth0:
-      dhcp4: false
-      addresses: ["{{ metal3_network_infra_public_ip }}/24"]
-      nameservers:
-        addresses: "{{ vm_egress_gw }}"
-      routes:
-        - to: default
-          via: "{{ vm_egress_gw }}"
-
 #
 # Public IPs
 #

diff --git a/playbooks/setup_metal3_network_infra.yml b/playbooks/setup_metal3_network_infra.yml
diff --git a/roles/diskimage_builder/.travis.yml b/roles/diskimage_builder/.travis.yml
diff --git a/roles/diskimage_builder/defaults/main.yml b/roles/diskimage_builder/defaults/main.yml