• pti-rtp [Jenkins] (o-ran-sc.org)

• pti-o2 [Jenkins] (o-ran-sc.org)

Open

Testing O2 IMS implementations across two platforms: OpenShift-KNI (PTI-RTP) and StarlingX (PTI-O2). The objective is validating baremetal provisioning workflows and managing real RAN workloads through O2 interfaces.

• Newton (Supermicro): Jumphost and O-Cloud Manager (OKD single-node)

• Galileo (Asus): Target for StarlingX O-Cloud Manager

• Joule (Dell R750): RT-capable node for gNB deployment via O2

• Laovisier (Supermicro): Reference gNB environment (RHEL 9.2, vanilla Kubernetes)

• Controller 00: 192.168.8.34

• Controller 00: 192.168.8.33

• Floating IP: 192.168.8.32

• Worker Node: TBD → either Joule or Lavoisier

StarlingX 10.0 on Galileo

Attempted simplex all-in-one installation. Bootstrap completed but Ceph storage backend installation fails consistently

StarlingX 8.0 on Galileo

Bootstrap fails during container image download phase. Upstream mirrors are missing required images. Error occurs during common/push-docker-images task:

• If I choose virtio as disk driver on cockpit instead of sata, the vg creation will be corrupt and there will be failure during bootstrap.

•  

TASK [bootstrap/prepare-env : Check volume groups] ******************************************************************************
Tuesday 25 November 2025 10:57:18 +0000 (0:00:00.024) 0:00:13.814 ******
fatal: [localhost]: FAILED! => changed=true
cmd:

• vgdisplay

• cgts-vg
  delta: '0:00:00.038214'
  end: '2025-11-25 10:57:19.038735'
  msg: non-zero return code
  rc: 5
  start: '2025-11-25 10:57:19.000521'
  stderr: |2-
  Volume group "cgts-vg" not found
  Cannot process volume group cgts-vg
  stderr_lines: <omitted>
  stdout: ''
  stdout_lines: <omitted>

PLAY RECAP **********************************************************************************************************************
localhost : ok=43 changed=8 unreachable=0 failed=1 skipped=78 rescued=0 ignored=0

common/prepare-env : stat ------------------------------------------------------------------------------------------------ 2.28s
common/prepare-env : Evaluate ansible_become_pass ------------------------------------------------------------------------ 0.89s
bootstrap/prepare-env : Create SSL CA cert directory --------------------------------------------------------------------- 0.71s
common/prepare-env : Append kubernetes extra configuration (extraArgs and extraVolumes) ---------------------------------- 0.69s
bootstrap/prepare-env : Check volume groups ------------------------------------------------------------------------------ 0.55s
bootstrap/prepare-env : Look for unmistakenly StarlingX packages --------------------------------------------------------- 0.53s
bootstrap/prepare-env : Check Docker status ------------------------------------------------------------------------------ 0.52s
bootstrap/prepare-env : Look for tmp kubernetes configuration file stats ------------------------------------------------- 0.52s
bootstrap/prepare-env : Clean up interface config files in /etc/network/interfaces.d/ ------------------------------------ 0.51s
bootstrap/prepare-env : Remove tmp kubernetes configuration file --------------------------------------------------------- 0.51s
common/validate-target : Retrieve software version number ---------------------------------------------------------------- 0.50s
common/validate-target : Retrieve system type ---------------------------------------------------------------------------- 0.50s
bootstrap/prepare-env : Check if bootstrap_finalized flag exists on host ------------------------------------------------- 0.50s
bootstrap/prepare-env : Look for openrc file ----------------------------------------------------------------------------- 0.49s
bootstrap/prepare-env : Look for last kubernetes configuration file stats ------------------------------------------------ 0.49s
bootstrap/prepare-env : Check initial config flag ------------------------------------------------------------------------ 0.49s
bootstrap/prepare-env : Fail if any of the mandatory configurations are not defined -------------------------------------- 0.12s
bootstrap/prepare-env : Copy the valid files referenced by the remaining overrides directly ------------------------------ 0.09s
bootstrap/prepare-env : Try to decode the content and write in the temp files if it's from a cert or key ----------------- 0.09s
bootstrap/prepare-env : Check if the files referenced by the remaining overrides exist ----------------------------------- 0.09s
sysadmin@localhost:~$ lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
loop0 7:0 0 256M 0 loop
-luks_encrypted_vault 252:4 0 240M 0 crypt /var/luks/stx/luks_fs sr0 11:0 1 3.1G 0 rom vda 253:0 0 500G 0 disk |-vda1 253:1 0 1M 0 part |-vda2 253:2 0 29.3G 0 part /var/rootdirs/opt/platform-backup |-vda3 253:3 0 300M 0 part /boot/efi |-vda4 253:4 0 2G 0 part /boot -vda5 253:5 0 468.4G 0 part
|-cgts--vg-root--lv 252:0 0 20G 0 lvm /sysroot
|-cgts--vg-var--lv 252:1 0 20G 0 lvm /var
|-cgts--vg-log--lv 252:2 0 7.8G 0 lvm /var/log
`-cgts--vg-scratch--lv 252:3 0 15.6G 0 lvm /var/rootdirs/scratch

Newton runs CentOS Stream 9 as the jumphost providing infrastructure services:

• domain: ocloud-vm-bmw.lab.local

Deployed using PTI-RTP repository (branch: l-release). The deployment creates a single-node OpenShift cluster (VM-AIO) on NAT network 192.168.123.0/24.

Network Architecture:

Since jumphost runs as VM behind a NAT network, need proxy to expose it towards outter network. 

Pre-built operator images (versions 4.18.0-4.21.0) fail to start with error:

This occurs because operator binaries were compiled against the old API group ocloud.openshift.io/v1alpha1, but CRDs in osc-l-release use o2ims.oran.openshift.io/v1alpha1.

• Built operator from source using osc-l-release branch

• We push ours here https://bmw.ece.ntust.edu.tw/repository/infidel/oran-o2ims

All operator components are now functional. API endpoints respond correctly to infrastructure inventory queries.

From this commit: https://gerrit.o-ran-sc.org/r/c/pti/rtp/+/15169

• I get them from here: https://github.com/openshift-kni/cnf-features-deploy.git

  • cnf-features-deploy/ztp/source-crs

Testing reveals a workflow failure during baremetal node provisioning. The sequence proceeds as follows:

1. Submit ProvisioningRequest via O2 API

2. Controller creates InfraEnv and BareMetalHost resources

3. Target machine boots inspection ISO via Redfish virtual media

4. Assisted Installer agent starts on target machine

5. Machine remains in inspection phase indefinitely

Open

Open

assisted-image-service is injected properly. All local domain names are reachable from CoreOS of target Baremetal

• Attempted to manually patch BareMetalHost resources to force transition to installation phase. This approach failed due to fundamental architectural issues:

• Patching BareMetalHost with installation ISO URL causes BMC to enter boot loop

• The inspection ISO (CoreOS-based) only runs ironic-agent and never starts agent.service

• Manually starting agent.service allows bootstrap to proceed, but the installation process writes CoreOS directly to disk, bypassing the intended cluster installation workflow

• domain: ocloud-vm-okd-aio.lab.localocloud-vm-okd-aio.la

• Installed

• Installed

• Need to rebuild o2ims too, default image complain about wrong CR

Baremetal Provisioning

• From this commit: https://gerrit.o-ran-sc.org/r/c/pti/rtp/+/15169

• BMH not creating anything

• assisted-installer-controller seems to be down

• │ time="2025-11-20T05:49:35Z" level=info msg="Start running Assisted-controller. Configuration is:\n struct ControllerConfig {\n\tClusterID: \"d4e36ee7-edd3-4de3-a70f-842 │ │ 21e884e62\",\n\tURL: \"http://192.168.123.11:8090/\",\n\tPullSecretToken: <SECRET>,\n\tSkipCertVerification: true,\n\tCACertPath: \"\",\n\tNamespace: \"assisted-install │ │ er\",\n\tOpenshiftVersion: \"4.19.0-okd-scos.0\",\n\tControlPlaneCount: 1,\n\tWaitForClusterVersion: false,\n\tMustGatherImage: \"quay.io/okd/scos-content@sha256:f738da │ │ cc63f9be039be41b5fa5f467b3f71cce070ea736dec2882f0722dad9d4\",\n\tDryRunEnabled: false,\n\tDryFakeRebootMarkerPath: \"\",\n\tDryRunClusterHostsPath: \"\",\n\tNotifyNumRe │ │ boots: true,\n\tParsedClusterHosts: config.DryClusterHosts(nil),\n}" func=main.main file="/go/src/github.com/openshift/assisted-installer/src/main/assisted-installer-co │ │ ntroller/assisted_installer_main.go:80" │ │ time="2025-11-20T05:49:35Z" level=info msg="Using proxy {HTTPProxy: HTTPSProxy: NoProxy:} to set env-vars for installer-controller pod" func="github.com/openshift/assis │ │ ted-installer/src/k8s_client.(*k8sClient).SetProxyEnvVars" file="/go/src/github.com/openshift/assisted-installer/src/k8s_client/k8s_client.go:425" │ │ time="2025-11-20T05:49:35Z" level=warning msg="Certificate verification is turned off. This is not recommended in production environments" func=github.com/openshift/ass │ │ isted-installer/src/inventory_client.CreateInventoryClientWithDelay file="/go/src/github.com/openshift/assisted-installer/src/inventory_client/inventory_client.go:93" │ │ time="2025-11-20T05:49:35Z" level=info msg="Making sure service dns-default can reserve the .10 address" func="github.com/openshift/assisted-installer/src/assisted_inst │ │ aller_controller.(*controller).HackDNSAddressConflict" file="/go/src/github.com/openshift/assisted-installer/src/assisted_installer_controller/assisted_installer_contro │ │ ller.go:310" │ │ time="2025-11-20T05:49:35Z" level=info msg="openshift-install-manifests ConfigMap attribute invoker = agent-installer" func=github.com/openshift/assisted-installer/src/ │ │ common.GetInvoker file="/go/src/github.com/openshift/assisted-installer/src/common/common.go:347" │ │ time="2025-11-20T05:49:35Z" level=warning msg="cluster is SNO and invoker = agent-installer, skipping assisted-installer-controller" func=main.main file="/go/src/github │ │ .com/openshift/assisted-installer/src/main/assisted-installer-controller/assisted_installer_main.go:157" │ │ stream closed: EOF for assisted-installer/assisted-installer-controller-6qm7w (assisted-installer-controller) │ │ │ │ │ │

Open

Open

• ironic-agent need these ports 6385, 6388, 5050. If you are behind a Proxy. Add them.

  • Wont’t work since Ironic will detect kvm IP everytime and updating the deployment cant be done since it will always reconcile

  • Curl from external network to haproxy ironic instance. Receive reponse but advertised link are KVM host link.

    • infidel ~ → curl -sk https://192.168.8.53:6385 | jq { "name": "OpenStack Ironic API", "description": "Ironic is an OpenStack project which enables the provision and management of baremetal machines.", "default_version": { "id": "v1", "links": [ { "href": "https://192.168.123.11:6385/v1/", "rel": "self" } ], "status": "CURRENT", "min_version": "1.1", "version": "1.95" }, "versions": [ { "id": "v1", "links": [ { "href": "https://192.168.123.11:6385/v1/", "rel": "self" } ], "status": "CURRENT", "min_version": "1.1", "version": "1.95" } ] } infidel ~ →

https://github.com/motangpuar/pti-rtp/commit/2c1a007f4979c80a9bd7ba177b7d0af36124db05

• Allow bridge attachment for master-node

• Mac and IP are dictacted by inventory templates, libvirt will generate machine templates accordingly

1. Groups templates

   1. --- ocloud_infra: vm ocloud_platform: okd ocloud_topology: aio ocloud_cluster_name: "ocloud-vm-bmw" ocloud_domain_name: "lab.local" ocloud_network_mode: "bridge" ocloud_net_bridge: "bridge0" ocloud_net_name: "br0-okd" ocloud_net_cidr: "192.168.8.0/24" ocloud_cluster_net_cidr: "10.128.0.0/14" ocloud_cluster_net_hostprefix: 23 ocloud_service_net_cidr: "172.30.0.0/16" ocloud_network_type: "OVNKubernetes" # DNS servers ocloud_dns_servers: - "192.168.8.72" # Optional NTP servers ocloud_ntp_servers: - "time.google.com"

2. Statically define IP and MAC inventory/host_vars/master-0-vm

   1. --- role: master ip_address: 192.168.8.210 ocloud_infra_vm_mem_gb: 48 mac_addresses: ens3: "52:54:00:ab:cd:01" network_config: interfaces: - name: ens3 type: ethernet state: up ipv4: enabled: true dhcp: false address: - ip: 192.168.8.210 prefix-length: 24 dns-resolver: config: server: - 192.168.8.72 routes: config: - destination: 0.0.0.0/0 next-hop-address: 192.168.8.103 next-hop-interface: ens3