Run the O-RAN Software Community hybrid M-Plane test runner (it-tifg) against a simulated
O-RAN radio unit on one COSMOS node. Any x86 server or node with 12 GB of RAM will do: this page
uses node1-1.sb5, a small 4-core desktop-class node. The O-RU calls home to an SMO-side O-RU controller over
TLS, and four conformance-style test cases check the connection, read the O-RU's data, count its
NETCONF sessions and write a configuration value through the controller. At the end you hold a
results file in the O-RAN test-results JSON format, and a framework you can extend with your own
test cases.
An O-RAN radio unit (O-RU) is managed over the Open Fronthaul M-Plane: a NETCONF server on the
O-RU, with configuration described by the O-RAN WG4 YANG models. In the hierarchical model only
the O-DU talks to the O-RU. In the hybrid model the SMO also manages the O-RU directly, through
an O-RU controller. Hybrid mode adds a second NETCONF client, a different way for the O-RU to
find its manager (NETCONF Call Home: the O-RU opens the connection to the controller, here over
TLS on port 4335), and new failure modes. The O-RAN WG4 conformance test specification describes
how to test it.
The O-RAN SC project it-tifg (integration and testing, test and integration focus group)
implements the first of those tests as a Python runner. It never speaks NETCONF itself. It drives
the O-RU through the controller's RESTCONF API, the way an SMO application would, and writes
its verdicts in the O-RAN test-results JSON schema. That design makes the runner independent of
the O-RU vendor, and it makes the controller part of what is tested.
This tutorial gives you the whole chain on one node with no radio: the O-RAN SC OAM solution
as the SMO-side controller (SDN-R on OpenDaylight, behind a traefik gateway, with Keycloak, Kafka
and a VES collector), two PyNTS network-element simulators (one O-RU, one O-DU), and it-tifg.
COSMOS packages all of it in the disk image hybrid-mplane.ndz, which brings the stack up at boot.
The same runner and controller are used against a real O-RU in a follow-on tutorial.
/rests/data/network-topology:network-topology/topology=topology-netconf/node=<name>.yang-ext:mount/ is the device's own YANG data, read and written as JSON.After completing this tutorial you will be able to:
it-tifg test cases checks and read its verdict in the O-RAN| Difficulty | Intermediate |
|---|---|
| Estimated time | About 5 min to load the image, 2 min to boot and bring the stack up, then about 70 s per test run |
| Domain / sandbox | Any: one x86 node with at least 12 GB of RAM. This page uses node1-1.sb5.cosmos-lab.org (Intel Core i7-4790, 4 cores, 15 GB) |
| Image | hybrid-mplane.ndz |
| Transmits? | No. The O-RU and O-DU are simulators |
| Tutorial package | cosmos-experiments-hybrid-mplane (cosmos-tutorial run hybrid-mplane) |
| Last verified | 2026-11-19: manual and Ansible modes on node1-1.sb5; the image also on an sb1 server |
Background knowledge
curl sends.Account & access
omf load, omf tell, omf stat: see OMF commands.Devices / nodes
| Resource | Role | Qty | Notes |
|---|---|---|---|
Any x86 node or server, e.g. node1-1.sb5.cosmos-lab.org |
SMO + simulators + test runner | 1 | At least 12 GB RAM (about 5.4 GB in use); the image fills 15 GB of disk. Needs no radio and no internet access at run time: every container image is in the disk image |
Disk images
| Image | Load onto | Provides |
|---|---|---|
hybrid-mplane.ndz |
the node | Ubuntu 24.04 (from baseline.ndz), Docker, the O-RAN SC OAM solution, PyNTS O-RU and O-DU images, it-tifg with the COSMOS patch, and the cosmos-hmp service that starts it all at boot |
Software components
| Component | Version | Source |
|---|---|---|
O-RAN SC OAM solution (SDN-R sdnr-image 13.0.1 on OpenDaylight netconf 8.0.7, traefik v3.3.6, Keycloak 26.1.2, Kafka, VES collector) |
o-ran-sc/oam commit 9b2dfe6 |
github.com/o-ran-sc/oam (solution/) |
PyNTS network-element simulators (pynts-o-ru-mplane, pynts-o-du-o1) |
0.9.5, o-ran-sc/sim-o1-ofhmp-interfaces commit ea08a2a |
github.com/o-ran-sc/sim-o1-ofhmp-interfaces |
Hybrid M-Plane test runner it-tifg |
commit ee6e413 plus two COSMOS patches (see How it works) |
github.com/o-ran-sc/it-tifg |
| Docker / Compose | Ubuntu docker.io 29.1, docker-compose-v2 2.40 |
Ubuntu 24.04 archive |
Spectrum / RF / special: none.
Everything runs on the one node. The O-RU simulator is the device under test; the controller is
the SMO side.
console.sb5.cosmos-lab.org
│ ssh root@node1-1
▼
node1-1.sb5 (hybrid-mplane.ndz) names: *.smo.o-ran-sc.org -> the node's CTRL address
┌────────────────────────────────────────────────────────────────────────────────────────────┐
│ it-tifg (Python, /opt/hmp/it-tifg) │
│ │ RESTCONF over HTTPS: https://controller.dcn.smo.o-ran-sc.org/rests/... │
│ ▼ │
│ gateway (traefik) ── :443 HTTPS ── :4335 TLS Call Home ── :4334 SSH Call Home │
│ │ ▲ │
│ ▼ │ O-RU opens the connection (Call Home) │
│ controller (SDN-R / OpenDaylight) │ │
│ │ NETCONF client sessions ──────────┼──────────────┐ │
│ ▼ │ ▼ │
│ pynts-o-ru-hybrid (O-RU, NETCONF server, O-RAN WG4 YANG) ◄── NETCONF ── pynts-o-du-o1 │
│ (O-DU simulator) │
│ identity (Keycloak) · kafka · ves-collector · odlux (web UI) · persistence · topology │
└────────────────────────────────────────────────────────────────────────────────────────────┘
smo.o-ran-sc.org andpynts-o-ru-hybrid.The names controller.dcn.smo.o-ran-sc.org, odlux.oam.smo.o-ran-sc.org and the rest are written
into the node's /etc/hosts at boot, pointing at its own CTRL address. Nothing outside the
node resolves them.
ssh <username>@console.sb5.cosmos-lab.org
omf tell -a offh -t node1-1
omf stat -t node1-1
-r 0) and allow a longomf load -i hybrid-mplane.ndz -t node1-1 -r 0 -o 2400 -b 600
Imaging Process Done
1 node successfully imaged - Topology saved in '/tmp/omf-load-1795089466-topo-success.rb'
Read the last line: 1 node successfully imaged is success. omf load exits 0 even on anode1-1.sb5.omf tell -a on -t node1-1
omf stat -t node1-1
cosmos-hmp service brings the stack up (about 80 s, after a 1 min 45 s boot onnode1-1.sb5). Follow it:ssh root@node1-1 'systemctl is-active cosmos-hmp; journalctl -u cosmos-hmp -b --no-pager -o cat | grep hmp-up'
active
[hmp-up 11:59:06Z] stopping network, oam and common stacks
[hmp-up 11:59:06Z] adopting /opt/hmp/oam/solution to <node-address> / smo.o-ran-sc.org
[hmp-up 11:59:06Z] starting common (gateway, identity, kafka, ...)
[hmp-up 11:59:55Z] creating users and identity realm
[hmp-up 11:59:58Z] starting oam (controller, odlux, ves-collector)
[hmp-up 12:00:25Z] controller healthy
[hmp-up 12:00:25Z] starting PyNTS O-DU (second NETCONF client for TC-003)
[hmp-up 12:00:26Z] it-tifg config written (0600)
[hmp-up 12:00:26Z] UP on <node-address>: controller https://controller.dcn.smo.o-ran-sc.org, ODLUX https://odlux.oam.smo.o-ran-sc.org
active and the line UP on <address> mean the controller is healthy and the O-DU simulator isactivating means it is still starting; ask again.The supported path. The tutorial ships as the cosmos-experiments-hybrid-mplane package on every
console, with its code in /usr/share/cosmos-experiments/hybrid-mplane/.
export OMF_NODES=node1-1
cosmos-tutorial check hybrid-mplane
cosmos-tutorial run hybrid-mplane --mode manual
## check 'hybrid-mplane' (/usr/share/cosmos-experiments/hybrid-mplane)
PASS subdomain sb5 is supported
...
PASS node1-1: the hybrid M-Plane rig (load hybrid-mplane.ndz)
PASS node1-1: docker running with the PyNTS images built
PASS node1-1: at least 12 GB RAM (the OAM controller, Kafka and Keycloak run together; ~5 GB in use)
PASS console has python3
PASS console has ansible-playbook
PASS console has ssh
PASS console has scp
RUNNABLE
...
## node node1-1 running image: hybrid-mplane.ndz
## 1. Wait for the boot-time bring-up (cosmos-hmp.service runs cosmos-hmp-up)
[hmp-up 12:00:25Z] controller healthy
[hmp-up 12:00:26Z] UP on <node-address>: controller https://controller.dcn.smo.o-ran-sc.org, ODLUX https://odlux.oam.smo.o-ran-sc.org
## 2. Containers on the node
controller Up 13 minutes (healthy)
...
## 3. Run the it-tifg test cases (001 002 003 004)
overall: PASS testId a021bd6b-1cac-4389-8d31-dff8169c288d
TC-001 PASS Transport and Handshake in IPv4/TLS Environment
TC-002 PASS Read-Only Data Retrieval (Unfiltered, Filtered, and Config-Only)
TC-003 PASS NETCONF Sessions Verification
TC-004 PASS O-RU Configurability Test (Positive Case)
[hmp-run 12:14:54Z] results: /root/hmp-results/20261119T121353Z
## 4. Fetch the results (/root/hmp-results/20261119T121353Z)
console.log
log_a021bd6b-1cac-4389-8d31-dff8169c288d_20261119T121353Z.log
pynts-log_a021bd6b-1cac-4389-8d31-dff8169c288d_20261119T121353Z.log
restconf-debug_a021bd6b-1cac-4389-8d31-dff8169c288d_20261119T121353Z.json
results_a021bd6b-1cac-4389-8d31-dff8169c288d_20261119T121353Z.json
## 5. Build the MANUAL-mode result contract
overall PASS: TC-001 PASS, TC-002 PASS, TC-003 PASS, TC-004 PASS
--mode ansible runs the same steps as an Ansible playbook and writes the same result contract
(hmp_contract_ansible.json). On node1-1.sb5 it ended with:
"msg": "overall PASS: TC-001 PASS, TC-002 PASS, TC-003 PASS, TC-004 PASS"
PLAY RECAP *********************************************************************
node1-1.sb5.cosmos-lab.org : ok=21 changed=5 unreachable=0 failed=0 skipped=1 rescued=0 ignored=0
cosmos-tutorial params hybrid-mplane lists what a --params my.yml overlay may change: which test
cases run (hmp_tests) and how long to wait for the bring-up (hmp_up_timeout_s).
The run directory (printed at the end) holds hmp_contract_manual.json (the verdict per test
case, the node, the image and its CPU platform), results/ with everything the runner wrote, and
run_record.json.
On the node, one command runs the four test cases:
ssh root@node1-1
cosmos-hmp-run
overall: PASS testId cff999d5-be76-4146-b3ab-f070a3b534db
TC-001 PASS Transport and Handshake in IPv4/TLS Environment
TC-002 PASS Read-Only Data Retrieval (Unfiltered, Filtered, and Config-Only)
TC-003 PASS NETCONF Sessions Verification
TC-004 PASS O-RU Configurability Test (Positive Case)
[hmp-run 12:18:55Z] results: /root/hmp-results/20261119T121753Z
cosmos-hmp-run 002 003 runs a subset. Each run first removes the O-RU container, because test
case 001 must see the O-RU call home after the runner has recorded the controller's state. Every
run writes a new /root/hmp-results/<UTC stamp>/ with passwords removed from the logs.
The controller's admin password is the upstream default, in /opt/hmp/oam/solution/smo/oam/.env
(ADMIN_PASSWORD). Read it into a variable instead of typing it:
PW=$(sed -n 's/^ADMIN_PASSWORD=//p' /opt/hmp/oam/solution/smo/oam/.env)
C=https://controller.dcn.smo.o-ran-sc.org/rests/data/network-topology:network-topology/topology=topology-netconf
curl -sk -u "admin:$PW" "$C?content=nonconfig" | python3 -m json.tool | grep -E '"node-id"|connection-status|"host"'
"node-id": "pynts-o-ru-hybrid",
"host": "172.19.0.2",
"connection-status": "connected",
The O-RU is the only mount point. The O-DU simulator is not managed by the controller: it is the
O-RU's second NETCONF client.
Read the O-RU's NETCONF sessions, the data test case 003 checks:
curl -sk -u "admin:$PW" "$C/node=pynts-o-ru-hybrid/yang-ext:mount/ietf-netconf-monitoring:netconf-state/sessions?content=nonconfig" | python3 -m json.tool
{
"ietf-netconf-monitoring:sessions": {
"session": [
{
"session-id": 3,
"in-rpcs": 23,
"login-time": "2026-11-19T12:14:13.068547387+00:00",
"out-rpc-errors": 1,
"in-bad-rpcs": 0,
"out-notifications": 2,
"source-host": "<node-address>",
"username": "smo",
"transport": "ietf-netconf-monitoring:netconf-tls"
},
{
"session-id": 1,
"in-rpcs": 25,
"login-time": "2026-11-19T12:14:08.030207539+00:00",
"out-rpc-errors": 0,
"in-bad-rpcs": 0,
"out-notifications": 3,
"source-host": "2001:db8:1:50::5",
"username": "hybrid-odu",
"transport": "ietf-netconf-monitoring:netconf-tls"
}
]
}
}
Two TLS sessions: smo is the controller (it arrives from the node's own address through the
gateway) and hybrid-odu is the O-DU simulator, over the Docker IPv6 network. That is the hybrid
M-Plane in one reply, and it is what TC-003 counts.
-k accepts the gateway's self-signed certificate. The O-RU stays connected after a run, so you
can keep exploring its tree under $C/node=pynts-o-ru-hybrid/yang-ext:mount/.
ODLUX, SDN-R's web UI, is served by the gateway on port 443 under the name
odlux.oam.smo.o-ran-sc.org, and its login goes through Keycloak at
identity.smo.o-ran-sc.org. The gateway routes by name, so the browser must use those names.
On your laptop:
/etc/hosts, orC:\Windows\System32\drivers\etc\hosts), pointing at your own machine:127.0.0.1 odlux.oam.smo.o-ran-sc.org identity.smo.o-ran-sc.org gateway.smo.o-ran-sc.org
sudo ssh -L 443:node1-1:443 <username>@console.sb5.cosmos-lab.org
https://odlux.oam.smo.o-ran-sc.org, accept the self-signed certificate (for both/opt/hmp/oam/solution/users.csv on the node (username and password columns).force_pwd_change, so the first sign-in asks you to choose a newpynts-o-ru-hybrid.Remove the hosts lines when you are done.
Bring-up (cosmos-hmp-up, at every boot). The image never trusts the address it was built
with. At boot the script resets the OAM solution to its pristine upstream files, runs upstream's
adopt_to_environment.py -d smo.o-ran-sc.org to write this node's CTRL address into the compose
environment and the O-RU's Call Home target, rewrites the smo.o-ran-sc.org block in /etc/hosts,
and then starts, in order: the common services (gateway, Keycloak, Kafka), the users and realm
(upstream create_users.py and config.py), the controller and ODLUX, and the O-DU simulator.
Finally it writes the runner's configuration with the controller's password (mode 0600). The
scripts are in /opt/hmp/cosmos/ on the node.
The test runner. it-tifg reads config/default.json, records the NETCONF connections the
controller already has, starts the O-RU simulator with Docker Compose, waits 60 s, and runs the
selected test cases. Each one talks only RESTCONF to the controller:
| Test case | What it checks | How |
|---|---|---|
| TC-001 Transport and Handshake in IPv4/TLS Environment | The O-RU called home over TLS and the controller accepted it | A new mount point appeared since the start, named pynts-o-ru-hybrid, connection-status connected, advertising the o-ran-uplane-conf capability, and the controller's Call Home status for it is CONNECTED over IPv4/TLS |
| TC-002 Read-Only Data Retrieval | The O-RU serves its data and the controller exposes it as JSON | Three GETs under the mount point: the whole tree, a subtree (ietf-hardware:hardware), and the same subtree as configuration only (content=config). Each must return 200 with data |
| TC-003 NETCONF Sessions Verification | Hybrid mode: the O-RU serves at least two NETCONF clients | Reads ietf-netconf-monitoring:netconf-state/sessions from the O-RU and needs two or more sessions: the controller's and the O-DU's |
| TC-004 O-RU Configurability Test (Positive Case) | The O-RU accepts a valid write made through the controller | PUTs o-ran-operations:operational-info with re-call-home-no-ssh-timer = 90 and max-call-home-attempts = 10, reads it back, and optionally looks for a netconf-config-change notification (not part of the verdict). The value is not restored afterwards |
The verdicts go to results_<testId>_<stamp>.json in the O-RAN test-results schema (contacts,
test lab, testbed components, test specification, then one result per test case with its
metrics), together with the runner log, the O-RU's own log and a RESTCONF request/response log.
The test-lab and testbed metadata are COSMOS's, in /opt/hmp/cosmos/metadata/pynts/.
What COSMOS changed, and why. All three upstream projects are used at pinned commits. These
changes were needed to make them work together on current software; each is small and in the
image's /opt/hmp/cosmos/ and /opt/hmp/patches/:
| Change | Why |
|---|---|
One it-tifg patch (branch cosmos-local): read the nested netconf-node-topology:netconf-node fields, accept the Call Home status key without its module prefix, drop a trailing slash from the mount URL, accept 201 from a PUT |
The runner was written against an older OpenDaylight. The controller in the OAM solution (netconf 8.0.7) nests the node fields and answers a create with 201 |
A controller health check override (docker-compose.cosmos-health.override.yaml) that probes RESTCONF |
sdnr-image 13.0.1 has no /ready endpoint, so the upstream health check never passes and traefik never routes to the controller |
A Docker drop-in, DOCKER_MIN_API_VERSION=1.24 |
traefik v3.3.6 speaks Docker API 1.24; Docker 29 refuses anything older than 1.44 by default |
traefik.docker.network labels on the controller (dcn) and the web services (dmz) (docker-compose.cosmos-network.override.yaml) |
The controller and ODLUX sit on more than one Docker network, and traefik picks one of their addresses at random. When it picks a network the gateway cannot reach, every RESTCONF call times out (504) after a reboot |
A second it-tifg patch: HMP_CALLHOME_TRIGGER |
Used only against a real O-RU, which cannot be restarted like a container: the command it names makes the O-RU call home after the runner has recorded the controller's state. Not used in this tutorial |
The test runner is by highstreet technologies and the O-RAN SC it-tifg project; the OAM solution
and PyNTS are O-RAN SC projects; all three are Apache-2.0.
The runner discovers any module named tc_hmp_<NNN>.py in its testcases package and takes the
test ID from the file name. A new test is one file. This example checks that the O-RU lists the
O-RAN U-Plane configuration module in its YANG schema list:
# /opt/hmp/it-tifg/src/hybrid_mplane_test_runner/testcases/tc_hmp_005.py
import logging
from datetime import datetime, UTC
import hybrid_mplane_test_runner.models.testresult_models as models
from hybrid_mplane_test_runner.config import get_config
from hybrid_mplane_test_runner.testcases.base import HybridMPlaneTestCase
class TestCase_005(HybridMPlaneTestCase):
categories = ["basic", "yang"]
suite = "COSMOS examples"
def __init__(self, controller=None):
super().__init__("005", "O-RU lists o-ran-uplane-conf in its YANG schemas",
"Reads ietf-netconf-monitoring schemas through the controller.",
controller=controller)
def run(self) -> models.TestCase:
self.tc_model.startDate = datetime.now(UTC)
name = get_config().get_config().test_cases.dut_mountpoint_name
url = (f"{self.controller.base_url}/rests/data/network-topology:network-topology/"
f"topology=topology-netconf/node={name}/yang-ext:mount/"
"ietf-netconf-monitoring:netconf-state/schemas?content=nonconfig")
status, data = self.controller._fetch_data(url)
schemas = (data or {}).get("ietf-netconf-monitoring:schemas", {}).get("schema", [])
found = any(s.get("identifier") == "o-ran-uplane-conf" for s in schemas)
logging.info(f"{len(schemas)} schemas, o-ran-uplane-conf listed: {found}")
ok = status == 200 and found
self.tc_model.result = models.ResultType.PASS if ok else models.ResultType.FAIL
self.tc_model.metrics = [models.Metric(
description="o-ran-uplane-conf in the O-RU schema list",
measurements=models.Measurements([
models.Measurement(name="Status Code", values=[str(status)], units=models.Units.text),
models.Measurement(name="Schema Count", values=[str(len(schemas))], units=models.Units.text),
]),
status=models.TestStatus.optional,
result=self.tc_model.result)]
self.tc_model.stopDate = datetime.now(UTC)
return self.get_test_case_model()
Run it alone, or with the others:
cosmos-hmp-run 005
cosmos-hmp-run 001 002 003 004 005
overall: PASS testId 29aa9d4d-0e99-4f69-815f-2c9b83c9ee90
TC-005 PASS O-RU lists o-ran-uplane-conf in its YANG schemas
[hmp-run 12:16:47Z] results: /root/hmp-results/20261119T121546Z
The file is gone at the next image load. Keep your own tests in your home directory and copy them
in after each load.
Write tests that read freely. A test that writes must restore what it changed: on a real O-RU the
next user's bring-up depends on its configuration (TC-004 is the upstream example of a write that
is not restored, which is harmless only on a simulator).
cosmos-hmp active, with UP on <CTRL address> in its journal.controller, odlux, ves-collector,pynts-o-du-o1, and, after a test run, pynts-o-ru-hybrid:controller Up 15 minutes (healthy)
gateway Up 16 minutes (healthy)
identity Up 16 minutes (healthy)
identitydb Up 16 minutes
kafka Up 16 minutes
kafka-bridge Up 16 minutes (healthy)
kafka-ui Up 15 minutes
odlux Up 15 minutes
persistence Up 16 minutes (healthy)
pynts-o-du-o1 Up 15 minutes
pynts-o-ru-hybrid Up About a minute
topology Up 16 minutes (healthy)
ves-collector Up 15 minutes (healthy)
zookeeper Up 16 minutes
overall: PASS with TC-001 to TC-004 each PASS, in hmp_contract_manual.json (overall,test_cases[]) and in the results JSON (testMetadata.result, testResults[].groupItems[].result).| Check | Command (on the node) | Expected |
|---|---|---|
| Bring-up finished | systemctl is-active cosmos-hmp |
active |
| Controller healthy | docker inspect -f '{{.State.Health.Status}}' controller |
healthy |
| Names resolve to the node | getent hosts controller.dcn.smo.o-ran-sc.org |
the node's CTRL address (<node-address> on node1-1.sb5) |
| Test run | cosmos-hmp-run |
overall: PASS, four PASS lines, exit status 0 |
| O-RU connected | the curl mount-point query above |
pynts-o-ru-hybrid with connection-status connected |
Results from the verification runs:
| CPU platform | Memory | Result | Run time | Measured |
|---|---|---|---|---|
| Intel Core i7-4790, 1 × 4 cores, 8 logical CPUs | 15 GB (5.4 GB in use) | TC-001..004 PASS, manual and Ansible modes; TC-005 example PASS | bring-up 80 s after boot; cosmos-hmp-run 61 s; cosmos-tutorial run 67 s (manual), 74 s (Ansible) |
2026-11-19, hybrid-mplane-20261119 |
| 2× Intel Xeon Gold 6126, 24 cores / 48 threads | 188 GB | TC-001..004 PASS | cosmos-hmp-run about 1 min |
2026-11-19, the rig the image was saved from |
Both platforms take the same time: the run is dominated by the runner's fixed 60 s wait for the
O-RU to call home, not by the CPU.
cosmos-tutorial run hybrid-mplane --mode teardown
omf tell -a offh -t node1-1
The teardown stops the cosmos-hmp service, which takes down the simulators, the controller and
the common services. Nothing needs saving: the next load of hybrid-mplane.ndz starts fresh.
| Symptom | Likely cause | Fix |
|---|---|---|
cosmos-hmp is failed |
A step of the bring-up failed; the journal names it | journalctl -u cosmos-hmp -b --no-pager -o cat, then systemctl restart cosmos-hmp (it starts from a clean state every time) |
controller not healthy after ... s |
The controller is slow to start, or Docker cannot reach it | docker logs controller --tail 50; restart the service. Allow 2 min |
traefik log: client version 1.24 is too old |
The Docker drop-in is missing | systemctl cat docker must show DOCKER_MIN_API_VERSION=1.24; it is in the image under /etc/systemd/system/docker.service.d/ |
| TC-001 FAIL, no new connection | The O-RU did not call home, or it called home before the runner recorded the controller's state | Run cosmos-hmp-run again: it removes the O-RU container first. Check getent hosts controller.dcn.smo.o-ran-sc.org gives the server's own address |
| TC-003 FAIL with one session | The O-DU simulator is not running | docker ps must list pynts-o-du-o1; systemctl restart cosmos-hmp |
curl returns 401 |
Wrong password variable | Re-read PW from /opt/hmp/oam/solution/smo/oam/.env as shown |
omf load exits 0 but the node does not boot the image |
The load timed out | Read the summary line; load again with -r 0 -o 2400 -b 600 |
omf stat shows POWEROFF right after a successful load |
Normal: the node is left off after imaging | omf tell -a on -t <node> |
| ODLUX sign-in asks for a new password | The users.csv users are created with force_pwd_change |
Choose a new password; it stays until the next image load |
On this wiki
External
solution/) with SDN-R.Author(s): COSMOS team · Last verified: 2026-11-19 (node1-1.sb5, manual and Ansible modes) · Tested image/release: hybrid-mplane-20261119 (runs above); hybrid-mplane.ndz now = hybrid-mplane-20261119b (same rig + the real-O-RU Call Home helpers, OS upgrade; load-back and bring-up verified on srv2-lg1.sb1 2026-11-19) (O-RAN SC oam 9b2dfe6, PyNTS 0.9.5, it-tifg ee6e413 + COSMOS patches; Ubuntu 24.04) · Tags: oran, m-plane, hybrid, netconf, call-home, restconf, sdn-r, o-ran-sc, it-tifg, pynts