This is the preferred and fast way to have your own local development RP setup, while also having a functional cluster. It uses hacks scripts around a lot of the setup to make things easier to bootstrap and be more sensible for running off of your local laptop.
- Check the specific use-case examples where deploying full RP service can be a better match.
- Your development environment is prepared according to the steps outlined in Prepare Your Dev Environment
-
Check the
env.example
file and copy it by creating your own:cp env.example env
-
Build the development
az aro
extension:. ./env make az
-
Verify the ARO extension is registered:
az -v ... Extensions: aro 0.4.0 (dev) /path/to/rp/python/az/aro ... Development extension sources: /path/to/rp/python ...
Note: you will be able to update your development
az aro
extension in the future by simply runninggit pull
. If you need to use the "prod" extension, what is bundled inaz
natively rather than your./python
, you canunset AZURE_EXTENSION_DEV_SOURCES
(found in your./env
file).
-
If you don't have access to a shared development environment and secrets, follow prepare a shared RP development environment.
-
If you have multiple subscriptions in your account, verify that "ARO SRE Team - InProgress (EA Subscription 2)" is your active subscription:
az account set --subscription "ARO SRE Team - InProgress (EA Subscription 2)"
-
Set SECRET_SA_ACCOUNT_NAME to the name of the storage account containing your shared development environment secrets and save them in
secrets
:SECRET_SA_ACCOUNT_NAME=rharosecretsdev make secrets
-
Run msi.sh to create a service principal and self-signed certificate to mock a cluster MSI. This script will also create the platform identities, platform identity role assignments, and role assignment on mock cluster MSI to federate the platform identities. Platform identities will be created in resource group
RESOURCEGROUP
and subscriptionSUBSCRIPTION
. Save the output values for cluster MSIClient ID
,Base64 Encoded Certificate
, andTenant
. Additionally, save the value forPlatform workload identity role sets
. -
Copy, edit (if necessary) and source your environment file. The required environment variable configuration is documented immediately below:
cp env.example env vi env . ./env
LOCATION
: Location of the shared RP development environment (default:eastus
).RP_MODE
: Set todevelopment
to use a development RP running at https://localhost:8443/.MOCK_MSI_CLIENT_ID
: Client ID for service principal that mocks cluster MSI (see previous step).MOCK_MSI_OBJECT_ID
: Object ID for service principal that mocks cluster MSI (see previous step).MOCK_MSI_CERT
: Base64 encoded certificate for service principal that mocks cluster MSI (see previous step).MOCK_MSI_TENANT_ID
: Tenant ID for service principal that mocks cluster MSI (see previous step).PLATFORM_WORKLOAD_IDENTITY_ROLE_SETS
: The platform workload identity role sets (see previous step or value inlocal_dev_env.sh
).
-
Create your own RP database:
az deployment group create \ -g "$RESOURCEGROUP" \ -n "databases-development-${AZURE_PREFIX:-$USER}" \ --template-file pkg/deploy/assets/databases-development.json \ --parameters \ "databaseAccountName=$DATABASE_ACCOUNT_NAME" \ "databaseName=$DATABASE_NAME" \ 1>/dev/null
-
Source your environment file.
. ./env
-
Run the RP
Option 1: using local compilation and binaries (requires local
go
/build dependencies/etc):make runlocal-rp
Option 2: using containerized build and run (requires local
podman
andopenvpn
):# establish a VPN connection to the shared dev environment Hive cluster sudo openvpn secrets/vpn-${LOCATION}.ovpn & # build/run the RP as a container make run-rp
-
To create a cluster, use one of the following methods:
NOTE: clusters created by a local dev RP will not be represented by Azure resources in ARM
-
Manually create the cluster using the public documentation.
Before following the instructions in Create, access, and manage an Azure Red Hat OpenShift 4 Cluster, you will need to manually register your subscription to your local RP:
curl -k -X PUT -H 'Content-Type: application/json' -d '{ "state": "Registered", "properties": { "tenantId": "'"$AZURE_TENANT_ID"'", "registeredFeatures": [ { "name": "Microsoft.RedHatOpenShift/RedHatEngineering", "state": "Registered" } ] } }' "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID?api-version=2.0"
Note that as long as the
RP_MODE
environment variable is set todevelopment
, theaz aro
client will connect to your local RP. -
Use the create utility:
# Create the cluster CLUSTER=<cluster-name> go run ./hack/cluster create
Later the cluster can be deleted as follows:
CLUSTER=<cluster-name> go run ./hack/cluster delete
By default, a public cluster will be created. In order to create a private cluster, set the
PRIVATE_CLUSTER
environment variable totrue
prior to creation. Internet access from the cluster can also be restricted by setting theNO_INTERNET
environment variable totrue
.
NOTE: If the cluster creation fails with
unable to connect to Podman socket...dial unix ///run/user/1000/podman/podman.sock: connect: no such file or directory
, then you will need enable podman user socket by executing :systemctl --user enable --now podman.socket
, and re-run the installation. -
-
Using
oc
:-
Get the KUBECONFIG:
az aro get-admin-kubeconfig \ --name <cluster-name> \ --resource-group <resource_group_name> \ --file dev.admin.kubeconfig; export KUBECONFIG=dev.admin.kubeconfig
The cluster and resource group names can be found by running
az aro list -o table
-
Setup insecure-tls
oc
:To interact with the cluster using using
oc
you will need to add the--insecure-skip-tls-verify
flag to all commands or use this handy alias:alias oc-dev='oc --insecure-skip-tls-verify'
Note: This is because the cluster is using self-signed certificates.
-
-
The following additional RP endpoints are available but not exposed via
az aro
:-
Delete a subscription, cascading deletion to all its clusters:
curl -k -X PUT \ -H 'Content-Type: application/json' \ -d '{"state": "Deleted", "properties": {"tenantId": "'"$AZURE_TENANT_ID"'"}}' \ "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID?api-version=2.0"
-
List operations:
curl -k \ "https://localhost:8443/providers/Microsoft.RedHatOpenShift/operations?api-version=2020-04-30"
-
View RP logs in a friendly format:
journalctl _COMM=aro -o json --since "15 min ago" -f | jq -r 'select (.COMPONENT != null and (.COMPONENT | contains("access"))|not) | .MESSAGE'
-
Optionally, create these aliases for viewing logs
cat >>~/.bashrc <<'EOF' alias rp-logs='journalctl _COMM=aro -o json --since "15 min ago" -f | jq -r '\''select (.COMPONENT != null and (.COMPONENT | contains("access"))|not) | .MESSAGE'\''' alias rp-logs-all='journalctl _COMM=aro -o json -e | jq -r '\''select (.COMPONENT != null and (.COMPONENT | contains("access"))|not) | .MESSAGE'\''' EOF
-
Sometimes you want to use a custom installer, for example, when you want to test a new OCP version's installer. You can create a cluster with the new installer following these steps:
-
Push the installer image to somewhere accessible from Hive AKS.
quay.io would be one of the options. You need pull-secret to use the repositories other than
arointsvc.azurecr.io
. It must be configured in the secrets. If you are using the hack script, you don't have to care about it because the script usesUSER_PULL_SECRET
automatically. -
Create a cluster with the version you updated.
If you are using the hack script, you can specify the version with
OS_CLUSTER_VERSION
env var.
If you are already familiar with running the ARO RP locally, you can speed up the process executing the local_dev_env.sh script.
The env variables names defined in pkg/util/liveconfig/manager.go control the communication of the ARO-RP with Hive.
- If you want to use ARO-RP + Hive, set
HIVE_KUBE_CONFIG_PATH
to the path of the kubeconfig of the AKS Dev cluster. Info about creating that kubeconfig (Step Access the cluster via API). - If you want to create clusters using the local ARO-RP + Hive instead of doing the standard cluster creation process (which doesn't use Hive), set
ARO_INSTALL_VIA_HIVE
to true. - If you want to enable the Hive adoption feature (which is performed during adminUpdate()), set
ARO_ADOPT_BY_HIVE
to true.
After setting the above environment variables (using export directly in the terminal or including them in the env file), connect to the VPN (Connect to the VPN section).
Then proceed to run the ARO-RP as usual.
After that, when you create a cluster, you will be using Hive behind the scenes. You can check the created Hive objects following Debugging OpenShift Cluster and using the oc command.
export CLUSTER=<cluster-name>
export AZURE_SUBSCRIPTION_ID=<subscription-id>
export RESOURCEGROUP=<resource-group-name>
[OR]
. ./env
-
Perform AdminUpdate on a dev cluster
curl -X PATCH -k "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER?api-version=admin" --header "Content-Type: application/json" -d "{}"
-
Get Cluster details of a dev cluster
curl -X GET -k "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER?api-version=admin" --header "Content-Type: application/json" -d "{}"
-
Get SerialConsole logs of a VM of dev cluster
VMNAME="aro-cluster-qplnw-master-0" curl -X GET -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/serialconsole?vmName=$VMNAME" --header "Content-Type: application/json" -d "{}"
-
Redeploy a VM in a dev cluster
VMNAME="aro-cluster-qplnw-master-0" curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/redeployvm?vmName=$VMNAME" --header "Content-Type: application/json" -d "{}"
-
Stop a VM in a dev cluster
VMNAME="aro-cluster-qplnw-master-0" curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/stopvm?vmName=$VMNAME" --header "Content-Type: application/json" -d "{}"
-
Stop and deallocate a VM in a dev cluster
VMNAME="aro-cluster-qplnw-master-0" curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/stopvm?vmName=$VMNAME&deallocateVM=True" --header "Content-Type: application/json" -d "{}"
-
Start a VM in a dev cluster
VMNAME="aro-cluster-qplnw-master-0" curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/startvm?vmName=$VMNAME" --header "Content-Type: application/json" -d "{}"
-
List VM Resize Options for a master node of dev cluster
curl -X GET -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/skus" --header "Content-Type: application/json" -d "{}"
-
Resize master node of a dev cluster
VMNAME="aro-cluster-qplnw-master-0" VMSIZE="Standard_D16s_v3" curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/resize?vmName=$VMNAME&vmSize=$VMSIZE" --header "Content-Type: application/json" -d "{}"
-
List Clusters of a local-rp
curl -X GET -k "https://localhost:8443/admin/providers/microsoft.redhatopenshift/openshiftclusters"
-
List cluster Azure Resources of a dev cluster
curl -X GET -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/resources"
-
Perform Cluster Upgrade on a dev cluster
curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/upgrade"
-
Get container logs from an OpenShift pod in a cluster
NAMESPACE=<namespace-name> POD=<pod-name> CONTAINER=<container-name> curl -X GET -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/kubernetespodlogs?podname=$POD&namespace=$NAMESPACE&container=$CONTAINER"
-
List Supported VM Sizes
VMROLE=<master or worker> curl -X GET -k "https://localhost:8443/admin/supportedvmsizes?vmRole=$VMROLE"
-
Perform Etcd Recovery Operation on a cluster
curl -X PATCH -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/etcdrecovery"
-
Delete a managed resource
MANAGED_RESOURCEID=<id of managed resource to delete> curl -X POST -k "https://localhost:8443/admin/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/deletemanagedresource?managedResourceID=$MANAGED_RESOURCEID"
-
We have a cosmos container which contains supported installable OCP versions, more information on the definition in
pkg/api/openshiftversion.go
. -
Admin - List OpenShift installation versions
curl -X GET -k "https://localhost:8443/admin/versions"
-
Admin - Put a new OpenShift installation version
This command adds the image to your cosmosDB. openShiftPullspec comes from quay.io/repository/openshift-release-dev ; and installerPullspec modifies as per example below, replace X.Y
accordingly.
curl -X PUT -k "https://localhost:8443/admin/versions" --header "Content-Type: application/json" -d '{ "properties": { "version": "4.14.16", "enabled": true, "openShiftPullspec": "quay.io/openshift-release-dev/ocp-release@sha256:XXXX", "installerPullspec": "arosvc.azurecr.io/aro-installer:release-X.Y"'
- List the enabled OpenShift installation versions within a region
curl -X GET -k "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID/providers/Microsoft.RedHatOpenShift/locations/$LOCATION/openshiftversions?api-version=2022-09-04"
-
Create a new OCM configuration
- You can find example payloads in the projects
./hack/ocm
folder.
curl -X PUT -k "https://localhost:8443/subscriptions/$AZURE_SUBSCRIPTION_ID/resourceGroups/$RESOURCEGROUP/providers/Microsoft.RedHatOpenShift/openShiftClusters/$CLUSTER/syncsets/mySyncSet?api-version=2022-09-04" --header "Content-Type: application/json" -d @./hack/ocm/syncset.b64
- You can find example payloads in the projects
-
SSH to the bootstrap node:
NOTE: If you have a password-based
sudo
command, you must first authenticate before runningsudo
in the backgroundsudo openvpn secrets/vpn-$LOCATION.ovpn & CLUSTER=cluster hack/ssh-agent.sh bootstrap
-
Get an admin kubeconfig:
CLUSTER=cluster make admin.kubeconfig export KUBECONFIG=admin.kubeconfig
-
"SSH" to a cluster node:
- Get the admin kubeconfig and
export KUBECONFIG
as detailed above. - Run the ssh-agent.sh script. This takes the argument is the name of the NIC attached to the VM you are trying to ssh to.
- Given the following nodes these commands would be used to connect to the respective node
$ oc get nodes NAME STATUS ROLES AGE VERSION aro-dev-abc123-master-0 Ready master 47h v1.19.0+2f3101c aro-dev-abc123-master-1 Ready master 47h v1.19.0+2f3101c aro-dev-abc123-master-2 Ready master 47h v1.19.0+2f3101c aro-dev-abc123-worker-eastus1-2s5rb Ready worker 47h v1.19.0+2f3101c aro-dev-abc123-worker-eastus2-php82 Ready worker 47h v1.19.0+2f3101c aro-dev-abc123-worker-eastus3-cbqs2 Ready worker 47h v1.19.0+2f3101c CLUSTER=cluster hack/ssh-agent.sh master0 # master node aro-dev-abc123-master-0 CLUSTER=cluster hack/ssh-agent.sh aro-dev-abc123-worker-eastus1-2s5rb # worker aro-dev-abc123-worker-eastus1-2s5rb CLUSTER=cluster hack/ssh-agent.sh eastus1 # worker aro-dev-abc123-worker-eastus1-2s5rb CLUSTER=cluster hack/ssh-agent.sh 2s5rb # worker aro-dev-abc123-worker-eastus1-2s5rb CLUSTER=cluster hack/ssh-agent.sh bootstrap # the bootstrap node used to provision cluster
- Get the admin kubeconfig and
- Connect to the VPN:
To access the cluster for oc / kubectl or SSH'ing into the cluster you need to connect to the VPN first.
NOTE: If you have a password-based
sudo
command, you must first authenticate before runningsudo
in the background
sudo openvpn secrets/vpn-aks-$LOCATION.ovpn &
-
Access the cluster via API (oc / kubectl):
make aks.kubeconfig export KUBECONFIG=aks.kubeconfig $ oc get nodes NAME STATUS ROLES AGE VERSION aks-systempool-99744725-vmss000000 Ready agent 9h v1.23.5 aks-systempool-99744725-vmss000001 Ready agent 9h v1.23.5 aks-systempool-99744725-vmss000002 Ready agent 9h v1.23.5
-
"SSH" into a cluster node:
- Run the ssh-aks.sh script, specifying the cluster name and the node number of the VM you are trying to ssh to.
hack/ssh-aks.sh aro-aks-cluster 0 # The first VM node in 'aro-aks-cluster' hack/ssh-aks.sh aro-aks-cluster 1 # The second VM node in 'aro-aks-cluster' hack/ssh-aks.sh aro-aks-cluster 2 # The third VM node in 'aro-aks-cluster'
-
Access via Azure Portal
Due to the fact that the AKS cluster is private, you need to be connected to the VPN in order to view certain AKS cluster properties, because the UI interrogates k8s via the VPN.
To run fake metrics socket:
go run ./hack/monitor
Steps to perform on Mac
- Mount your local MacOS filesystem into the podman machine:
podman machine init --now --cpus=4 --memory=4096 -v $HOME:$HOME
- Use the openvpn config file (which is now mounted inside the podman machine) to start the VPN connection:
podman machine ssh
sudo rpm-ostree install openvpn
sudo systemctl reboot
podman machine ssh
sudo openvpn --config /Users/<user_name>/go/src/github.com/Azure/ARO-RP/secrets/vpn-aks-westeurope.ovpn --daemon --writepid vpnpid
ps aux | grep openvpn
Update the env File
- Open the
env
file. - Update env file instructions: set
OPENSHIFT_VERSION
, updateINSTALLER_PULLSPEC
andOCP_PULLSPEC
, mention quay.io for SHA256 hash. - Update INSTALLER_PULLSPEC with the appropriate name and tag, typically matching the OpenShift version, e.g.,
release-4.13.
(for more detail see theenv.example
)
- Source the environment file before creating the cluster using the
setup_resources.sh
script(Added the updated env in the PR)
cd /hack
./setup_resources.sh
- Once the cluster create verify connectivity with the ARO cluster:
- Download the admin kubeconfig file
az aro get-admin-kubeconfig --name <cluster_name> --resource-group v4-westeurope --file ~/.kube/aro-admin-kubeconfig
- Set the KUBECONFIG environment variable
export KUBECONFIG=~/.kube/aro-admin-kubeconfig
- Verify connectivity with the ARO cluster
kubectl get nodes
kubectl get nodes
NAME STATUS ROLES AGE VERSION
shpaitha-aro-cluster-4sp5c-master-0 Ready control-plane,master 39m v1.25.11+1485cc9
shpaitha-aro-cluster-4sp5c-master-1 Ready control-plane,master 39m v1.25.11+1485cc9
shpaitha-aro-cluster-4sp5c-master-2 Ready control-plane,master 39m v1.25.11+1485cc9
shpaitha-aro-cluster-4sp5c-worker-westeurope1-j9c76 Ready worker 29m v1.25.11+1485cc9
shpaitha-aro-cluster-4sp5c-worker-westeurope2-j9zrs Ready worker 27m v1.25.11+1485cc9
shpaitha-aro-cluster-4sp5c-worker-westeurope3-56tk7 Ready worker 28m v1.25.11+1485cc9
- Trying to use
az aro
CLI in Production, fails with:
(NoRegisteredProviderFound) No registered resource provider found for location '$LOCATION' and API version '2024-08-12-preview'
-
Check if
~/.azure/config
there is a blockextensions.dev_sources
. If yes, comment it. -
Check if env var
AZURE_EXTENSION_DEV_SOURCES
is set. If yes, unset it. -
Installation fails with authorization errors:
Message="authorization.RoleAssignmentsClient#Create: Failure responding to request: StatusCode=403 -- Original Error: autorest/azure: Service returned an error. Status=403 Code=\"AuthorizationFailed\" Message=\"The client '$SP_ID' with object id '$SP_ID' does not have authorization to perform action 'Microsoft.Authorization/roleAssignments/write' over scope '/subscriptions/$SRE_SUBSCRIPTION/resourceGroups/$myresourcegroup/providers/Microsoft.Authorization/roleAssignments/b5a083aa-f555-466e-a268-4352b3b8394d' or the scope is invalid. If access was recently granted, please refresh your credentials.\"" Target="encountered error"
exit status 1
To resolve, check if it has the User Access Administrator
role assigned.
az role assignment list --assignee $SP_ID --output json --query '[].{principalId:principalId, roleDefinitionName:roleDefinitionName, scope:scope}'