TESTBenches.md

Table of Contents

• Machines
• Testbenches
  • 5G OTA Testbench
  • 5G NSA/Faraday Cage Testbench
  • 5G AW2S Testbench
  • 5G UE OTA Testbench
  • 4G Testbench(es)
• Pipelines
  • RAN-Container-Parent
    • Image Build pipelines
    • Image Test pipelines
  • RAN-CI-NSA-Trigger
• How to reproduce CI results
• How to debug CI failures
  • Developer instructions
  • CI team instructions
    • Core dump in a file
    • Core dump via systemd
    • Core dump via abrt (automatic bug reporting tool)
    • Core dump via apport

Machines

Machine	IP address	Lockable Resource	Function	Connected devices
asterix	172.21.16.127	CI-Asterix-Usage	unused	172.21.19.14
obelix	172.21.16.128	CI-Obelix-Usage	eNB (n40, n78), nrUE	172.21.19.13, X300 (192.168.60.2)
porcepix	172.21.16.136	CI-Porcepix	Executor, EPC, 5GC	--
up2	172.21.19.68	CI-UP2-Usage	COTS UE	Quectel RM520N
nepes	172.21.16.137	CI-Nepes	gNB (n78), EPC/5GC	B200mini (30C51EB)
ofqot	172.21.16.109	CI-Ofqot	gNB (n78)	B200mini (30C51D4)
idefix	172.21.16.135	CI-Idefix	COTS UE	Quectel RM500Q
caracal	172.21.16.132	CI-Caracal	gNB/phytest	N300 (192.168.10.2)
amariue	172.21.16.144	CI-Amarisoft-UE-Usage	nrUE	Amarisoft UE simulator
nano	172.21.18.48	CI-Nano-Legacy-EPC	Executor, EPC, adb	2x COTS (adb)
hutch	172.21.18.46	CI-Hutch-Legacy-FDD-eNB	eNB (B7)	B200mini (30C5239)
starsky	172.21.18.45	CI-Starsky-Legacy-TDD-eNB	eNB (B40)	b200mini (30A3E3C)
carabe	172.21.18.47	CI-Carabe-Legacy-FDD-OAI-LTE-UE	UE (B7UE)	B200mini (30AE8C9)
nokiabox	172.21.19.39	None	gNB (Nokia), 5GC	Nokia RF integrated
avra	172.21.16.124	CI-Avra-Usage	gNB (n78)	AW2S Jaguar (192.168.80.239)
orion	172.21.16.134	CI-Orion-Build-Sanity-Check-Deploy-Test, CI-Orion-DsTester-Deploy-Test	Build
aerial1	172.21.16.130	CI-Aerial2-Usage	gNB (PNF/Nvidia CUBB + VNF)	Foxconn RU, Nvidia Aerial SDK integrated
sphex	172.21.17.54	CI-Sphex-Usage	COTS UE	Quectel RM500Q
matix	172.21.19.58	CI-Matix-Usage	gNB (n77)	N310

Note: The available resources, and their current usage, is indicated here:

• Lockable resources of jenkins-oai: "New" Jenkins, i.e., with RAN-Container-Parent

Testbenches

5G OTA Testbench

Purpose: Over-the-air 4G/5G (NSA/SA) tests, performance tests

PDF version | LaTeX/TikZ version if you want to modify to reflect your setup

5G NSA/Faraday Cage Testbench

Purpose: Faraday cage 5G tests, functional tests

PDF version | LaTeX/TikZ version if you want to modify to reflect your setup

5G AW2S Testbench

Purpose: AW2S tests with Amarisoft UE simulator

PDF version | LaTeX/TikZ version if you want to modify to reflect your setup

5G UE OTA Testbench

Purpose: Over-the-air 5G tests with OAI UE

PDF version | LaTeX/TikZ version if you want to modify to reflect your setup

4G Testbench(es)

Purpose: 4G/LTE testbenches

PDF version | LaTeX/TikZ version if you want to modify to reflect your setup

Pipelines

RAN-Container-Parent

Purpose: automatically triggered tests on MR creation or push, from Gitlab Webhook ~documentation ~BUILD-ONLY ~4G-LTE ~5G-NR

This pipeline has basically two main stages, as follows. For the image build, please also refer to the dedicated documentation for information on how the images are built.

Image Build pipelines

• RAN-ARM-Cross-Compile-Builder ~BUILD-ONLY ~4G-LTE ~5G-NR
  • orion: Cross-compilation from Intel to ARM
  • base image from Dockerfile.base.ubuntu20.cross-arm64
  • build image from Dockerfile.build.ubuntu20.cross-arm64 (no target images)
• RAN-cppcheck ~BUILD-ONLY ~4G-LTE ~5G-NR
  • bellatrix
  • performs static code analysis, currently not actively enforced
• RAN-RHEL8-Cluster-Image-Builder ~BUILD-ONLY ~4G-LTE ~5G-NR
  • cluster (Asterix-OC-oaicicd-session resource): RHEL image build using the OpenShift Cluster (using gcc/clang)
  • base image from Dockerfile.build.rhel9
  • build image from Dockerfile.build.rhel9, followed by
    • target image from Dockerfile.eNB.rhel9
    • target image from Dockerfile.gNB.rhel9,
    • target image from Dockerfile.gNB.aw2s.rhel9
    • target image from Dockerfile.nr-cuup.rhel9
    • target image from Dockerfile.lteUE.rhel9
    • target image from Dockerfile.nrUE.rhel9
  • build image from Dockerfile.phySim.rhel9 (creates as direct target physical simulator image)
  • build image from Dockerfile.clang.rhel9 (compilation only, artifacts not used currently)
• RAN-Ubuntu18-Image-Builder ~BUILD-ONLY ~4G-LTE ~5G-NR
  • run formatting check from ci-scripts/docker/Dockerfile.formatting.bionic
  • obelix: Ubuntu 20 image build using docker (Note: builds U20 images while pipeline is named U18!)
  • base image from Dockerfile.base.ubuntu20
  • build image from Dockerfile.build.ubuntu20, followed by
    • target image from Dockerfile.eNB.ubuntu20
    • target image from Dockerfile.gNB.ubuntu20
    • target image from Dockerfile.nr-cuup.ubuntu20
    • target image from Dockerfile.nrUE.ubuntu20
    • target image from Dockerfile.lteUE.ubuntu20
    • target image from Dockerfile.lteRU.ubuntu20
  • build unit tests from ci-scripts/docker/Dockerfile.unittest.ubuntu20, and run them

Image Test pipelines

• OAI-CN5G-COTS-UE-Test ~5G-NR
  • using 5GC bench (resources CI-Cetautomatix-OC-oaicicd-session, CI-Dogmatix-CN5G-gNB): Attach/Detach of UE with multiple PDU sessions
• RAN-gNB-N300-Timing-Phytest-LDPC ~5G-NR
  • caracal + N310
  • pure performance test through phy-test scheduler, see command line for more details
• RAN-Interop-F1 ~5G-NR
  • ofqot (DU, 1x UE)
  • F1 interoperability: set up connection between Accelleran CU and OAI DU and pass all traffic over F1
  • 3rd-party gNB/CU interoperability: set up connection between Accelleran CU and OAI UE and test connectivity
• RAN-L2-Sim-Test-4G ~4G-LTE
  • obelix (eNB, 1x UE, OAI EPC)
  • L2simulator: skips physical layer and uses proxy between eNB and UE
• RAN-L2-Sim-Test-5G ~5G-NR
  • obelix (gNB, 1x UE, OAI 5GC)
  • L2simulator: skips physical layer and uses proxy between gNB and UE, currently only ping
• RAN-LTE-FDD-LTEBOX-Container ~4G-LTE
  • hutch + B210, nano w/ ltebox + 2x UE
  • tests RRC inactivity timers, different bandwidths, IF4p5 fronthaul
• RAN-LTE-FDD-OAIUE-OAICN4G-Container ~4G-LTE
  • hutch + B210 (eNB), carabe + B210 (4G UE), nano w/ OAI 4GC
  • tests OAI 4G for 10 MHz/TM1; known to be unstable
• RAN-LTE-TDD-2x2-Container ~4G-LTE
  • obelix + N310, porcepix, up2 + Quectel
  • TM1 and TM2 test, IF4p5 fronthaul
• RAN-LTE-TDD-LTEBOX-Container ~4G-LTE
  • starsky + B210, nano w/ ltebox + 2x UE
  • TM1 over bandwidths 5, 10, 20 MHz in Band 40, default scheduler for 20 MHz
• RAN-NSA-B200-Module-LTEBOX-Container ~4G-LTE ~5G-NR
  • nepes + B200 (eNB), ofqot + B200 (gNB), idefix + Quectel, nepes w/ ltebox
  • basic NSA test
• RAN-PhySim-Cluster ~4G-LTE ~5G-NR
  • cluster (Asterix-OC-oaicicd-session resource), tests in OpenShift Cluster
  • unitary simulators (nr_dlsim, etc.)
  • see ./physical-simulators.md for an overview
• RAN-RF-Sim-Test-4G ~4G-LTE
  • obelix (eNB, lteUE, OAI EPC)
  • uses RFsimulator, for FDD 5, 10, 20MHz with core, 5MHz noS1
• RAN-RF-Sim-Test-5G ~5G-NR
  • obelix (gNB, nrUE, OAI 5GC)
  • uses RFsimulator, TDD 40MHz, FDD 40MHz, F1 split
• RAN-SA-AW2S-CN5G ~5G-NR
  • 5G-NR SA test setup: avra(RHEL9.1) + AW2S, amariue, OAI CN5G
  • uses OpenShift cluster for CN deployment and container images for gNB deployment
  • multi UE testing using Amarisoft UE simulator
• RAN-SA-B200-Module-SABOX-Container ~5G-NR
  • ofqot + B200, idefix + Quectel, nepes w/ sabox
  • basic SA test (20 MHz TDD), F1, reestablishment, ...
• RAN-SA-OAIUE-CN5G ~5G-NR
  • 5G-NR SA test setup: gNB on avra(RHEL9.2) + N310, OAIUE on caracal(RHEL9.1) + N310, OAI CN5G
  • OpenShift cluster for CN deployment and container images for gNB and UE deployment
• RAN-SA-AERIAL-CN5G ~5G-NR
  • 5G-NR SA test setup: OAI VNF + PNF/NVIDIA CUBB on Aerial1 (U22) + Foxconn RU, sphex + COTS UE (Quectel RM500Q), OAI CN5G
  • container images for gNB deployment
• RAN-SA-2x2-Module-CN5G ~5G-NR
  • matix + N310 (gNB), up2 + COTS UE (Quectel RM520N), OAI 5GC deployed in docker on matix
  • NR performance tests: 2x2 configuration, 60 MHz and 100 MHz bandwidth

RAN-CI-NSA-Trigger

DEFUNCT: longer-running over-the-air LTE, NSA, and SA tests. To be integrated into RAN-Container-Parent.

• RAN-NSA-2x2-Module-OAIEPC
  • obelix + N310 (eNB), asterix + N310 (gNB), nrmodule2 + Quectel, porcepix w/ Magma EPC
  • LTE 2x2 and NR 2x2 (non-standalone)

How to reproduce CI results

The CI builds docker images at the beginning of every test run. To see the exact command line steps, please refer to the docker/Dockerfile.build.* files. Note that the console log of each pipeline run also lists the used docker files.

The CI uses these images for most of the pipelines. It uses docker-compose to orchestrate the images. To identify the docker-compose file, follow these steps:

1. Each CI test run HTML lists the XML file used for a particular test run. Open the corresponding XML file under ci-scripts/xml_files/.
2. The XML file has a "test case" that refers to deployment of the image; it will reference a directory containing a YAML file (the docker-compose file) using option yaml_path, which will be under ci-scripts/yaml_files/. Go to this directory and open the docker-compose file.
3. The docker-compose file can be used to run images locally, or you can infer the used configuration file and possible additional options that are to be passed to the executable to run from source.

For instance, to see how the CI runs the multi-UE 5G RFsim test case, the above steps look like this:

1. The first tab in the 5G RFsim test mentions xml_files/container_5g_rfsim.xml, so open ci-scripts/xml_files/container_5g_rfsim.xml.
2. This XML file has a DeployGenObject test case, referencing the directory yaml_files/5g_rfsimulator. The corresponding docker-compose file path is ci-scripts/yaml_files/5g_rfsimulator/docker-compose.yaml.
3. To know how to run the gNB, realize that there is a section oai-gnb. It mounts the configuration ci-scripts/conf_files/gnb.sa.band78.106prb.rfsim.conf (note that the path is relative to the directory in which the docker-compose file is located). Further, an environment variable USE_ADDITIONAL_OPTIONS is declared, referencing the relevant options --sa -E --rfsim (you can ignore logging options). You would therefore run the gNB from source like this:

   sudo ./cmake_targets/ran_build/build/nr-softmodem -O ci-scripts/conf_files/gnb.sa.band78.106prb.rfsim.conf --sa -E --rfsim

   To run this on your local machine, assuming you have a 5GC installed, you might need to change IP information in the config to match your core.

If you wish, you can rebuild CI images locally following these steps and then use the docker-compose file directly.

Some tests are run from source (e.g. ci-scripts/xml_files/gnb_phytest_usrp_run.xml), which directly give the options they are run with.

How to debug CI failures

It is possible to debug CI failures using the generated core dump and the image used for the run. A script is provided (see developer instructions below) that, provided the core dump file, container image, and the source tree, executes gdb inside the container; using the core dump information, a developer can investigate the cause of failure.

Developer instructions

The CI team will send you a docker image and a core dump file, and the commit as of which the pipeline failed. Let's assume the coredump is stored at /tmp/coredump.tar.xz, and the image is in /tmp/oai-nr-ue.tar.gz. First, you should check out the corresponding branch (or directly the commit), let's say in ~/oai-branch-fail. Now, unpack the core dump, load the image into docker, and use the script docker/debug_core_image.sh to open gdb, as follows:

cd /tmp
tar -xJf /tmp/coredump.tar.xz
docker load < /tmp/oai-nr-ue.tar.gz
~/oai-branch-fail/docker/debug_core_image.sh <image> /tmp/coredump ~/oai-branch-fail

where you replace <image> with the image loaded in docker load. The script will start the container and open gdb; you should see information about where the failure (e.g., segmentation fault) happened. If you just see ??, the core dump and container image don't match. Be also on the lookout for the corresponding message from gdb:

warning: core file may not match specified executable file.

Once you quit gdb, the container image will be removed automatically.

CI team instructions

The entrypoint scripts of all containers print the core pattern that is used on the running machine. Search for core_pattern at the start of the container logs to retrieve the possible location. Possible locations might be:

• a path: the corresponding directory must be mounted in the container to be writable
• systemd-coredumpd: see documentation
• abrt: see documentation
• apport: see documentation

See below for instructions on how to retrieve the core dump. Further, download the image and store it to a file using docker save. Make sure to pick the right image (Ubuntu or RHEL)!

Core dump in a file

This is not recommended, as files could pile up and fill the system disk completely! Prefer another method further down.

If the core pattern is a path: it should at least include the time in the pattern name (suggested pattern: /tmp/core.%e.%p.%t) to correlate the time the segfault occurred with the CI logs. If you identified the core dump, copy the core dump from that machine; if identification is difficult, consider rerunning the pipeline.

Core dump via systemd

Use the first command to list all core dumps. Scroll down to the core dump of interest (it lists the executables in the last column; use the time to correlate the segfault and the CI run). Take the PID of the executable (first column after the time). Dump the core dump to a location of your choice.

sudo coredumpctl list
sudo coredumpctl dump <PID> > /tmp/coredump

Core dump via abrt (automatic bug reporting tool)

TBD: use the documentation page for the moment.

Core dump via apport

I did not find an easy way to use apport. Anyway, the systemd approach works fine. So remove apport, install systemd-coredump, and verify it is the new coredump handler:

sudo systemctl stop apport
sudo systemctl mask --now apport
sudo apt install systemd-coredump
# Verify this changed the core pattern to a pipe to systemd-coredump
sysctl kernel.core_pattern