Ok you’re missing the kernel driver that DPDK needs to work on your host. Since Trex cannot find any suitable driver it tries to build one (hence make complaining ity can't find tools to recompile in the container).
http://doc.dpdk.org/guides/linux_gsg/linux_drivers.html You have the choice of few drivers, we use vfio for nfvbench so you may want to install and try that one first. Alec From: "Li, Charlie" <charlie...@amd.com> Date: Monday, August 31, 2020 at 4:35 PM To: "Alec Hothan (ahothan)" <ahot...@cisco.com>, "opnfv-tech-discuss@lists.opnfv.org" <opnfv-tech-discuss@lists.opnfv.org> Subject: RE: [opnfv-tech-discuss] #nfvbench - Example of run fails [AMD Public Use] Hi Alec, I am using Ubuntu 20.04.1 with kernel 5.4.0-42. Hugepage is configured in the grub $ grep HugePages_ /proc/meminfo HugePages_Total: 4096 HugePages_Free: 4096 HugePages_Rsvd: 0 HugePages_Surp: 0 In the host, I can manually bind the devices to vfio-pci, but in the container, it cannot bind the devices to vfio-pci or igb_uio. Please see the logs in the attached file. Regards, Charlie Li From: Alec Hothan (ahothan) <ahot...@cisco.com> Sent: Monday, August 31, 2020 5:36 PM To: Li, Charlie <charlie...@amd.com>; opnfv-tech-discuss@lists.opnfv.org Subject: Re: [opnfv-tech-discuss] #nfvbench - Example of run fails [CAUTION: External Email] This likely indicates that Trex failed to start properly, possibly pci address issue or other dpdk related issue. Can you check/share the trex config and trex log which are located inside the container at: /etc/trex_cfg.yaml /tmp/trex.log (you can use docker cp or simply docker exec <nfvbench> cat /tmp/trex.log) Make sure you have your server is set properly to run dpdk (you need huge pages configured). Alec From: "Li, Charlie" <charlie...@amd.com<mailto:charlie...@amd.com>> Date: Monday, August 31, 2020 at 3:07 PM To: "Alec Hothan (ahothan)" <ahot...@cisco.com<mailto:ahot...@cisco.com>>, "opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>" <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: RE: [opnfv-tech-discuss] #nfvbench - Example of run fails [AMD Public Use] Thanks Alec for your help. Now I can start it but it fails to connect to the server. $ nfvbench -c /tmp/nfvbench/nfvbench.cfg --rate 10kpps 2020-08-31 21:43:30,092 INFO Loading configuration file: /tmp/nfvbench/nfvbench.cfg 2020-08-31 21:43:30,101 INFO -c /tmp/nfvbench/nfvbench.cfg --rate 10kpps 2020-08-31 21:43:30,104 INFO Connecting to TRex (127.0.0.1)... 2020-08-31 21:43:33,175 INFO Starting TRex ... 2020-08-31 21:43:33,178 INFO nohup /bin/bash -c ./t-rex-64 -i -c 4 --iom 0 --no-scapy-server --unbind-unused-ports --close-at-end --vlan --hdrh --mbuf-factor 0.2 --cfg /etc/trex_cfg.yaml &> /tmp/trex.log & disown 2020-08-31 21:43:33,181 INFO TRex server is running... nohup: ignoring input and appending output to 'nohup.out' 2020-08-31 21:43:37,189 INFO Retrying connection to TRex (*** [RPC] - Failed to get server response from tcp://127.0.0.1:4501)... 2020-08-31 21:43:41,196 INFO Retrying connection to TRex (*** [RPC] - Failed to get server response from tcp://127.0.0.1:4501)... 2020-08-31 21:43:45,203 INFO Retrying connection to TRex (*** [RPC] - Failed to get server response from tcp://127.0.0.1:4501)... I tried to bind the devices to vfio-pci manually but it still cannot connect to the server. Regards, Charlie Li From: Alec Hothan (ahothan) <ahot...@cisco.com<mailto:ahot...@cisco.com>> Sent: Monday, August 31, 2020 1:45 PM To: Li, Charlie <charlie...@amd.com<mailto:charlie...@amd.com>>; opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org> Subject: Re: [opnfv-tech-discuss] #nfvbench - Example of run fails [CAUTION: External Email] Hi Charlie, This is likely due to the file "/tmp/nfvbench/nfvbench.cfg" missing inside the container. Can you check the following: · Check where you stored the nfvbench.cfg file on the host (the quickstart guide suggests to put the file under $HOME/nfvbench) · Check that that directory on host is mapped to /tmp/nfvbench in the container To verify you can also try to cat that file from inside the container (docker exec <container name> cat /tmp/nfvbench/nfvbench.cfg) For further interaction, you can continue on this email thread or you can join opnfv.slack.com #nfvbench if you prefer. Thanks Alec From: <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> on behalf of "Li, Charlie" <charlie...@amd.com<mailto:charlie...@amd.com>> Date: Monday, August 31, 2020 at 9:59 AM To: "opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>" <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: [opnfv-tech-discuss] #nfvbench - Example of run fails [AMD Public Use] Hello, I am new to NFVbench and is following the link below to get started https://opnfv-nfvbench.readthedocs.io/en/latest/testing/user/userguide/quickstart_docker.html<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fopnfv-nfvbench.readthedocs.io%2Fen%2Flatest%2Ftesting%2Fuser%2Fuserguide%2Fquickstart_docker.html&data=02%7C01%7CCharlie.Li%40amd.com%7Ca8b345a1f5134d2f91f508d84dfe3b03%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637345101599997618&sdata=Yh8wvZoPBaBXqCGaHLZbZvidWTg%2B%2BFQvqfG1TJ9x3vg%3D&reserved=0> The NFVbench container is started successfully $ nfvbench --version 4.3.1.dev4 But when I try the example of run, it fails $ nfvbench -c /tmp/nfvbench/nfvbench.cfg --rate 10kpps 2020-08-27 20:10:45,424 INFO Loading configuration string: /tmp/nfvbench/nfvbench.cfg 2020-08-27 20:10:45,425 ERROR String /tmp/nfvbench/nfvbench.cfg is not well formatted. Please verify your yaml/json string. If string is a file path, file was not found. Please use correct path and verify it is visible to container if you run nfvbench in container. 2020-08-27 20:10:45,425 ERROR {'status': 'ERROR', 'error_message': 'Traceback (most recent call last):\n File "/nfvbench/nfvbench/config.py", line 46, in config_loads\n cfg = AttrDict(yaml.safe_load(cfg_text))\n File "/usr/local/lib/python3.6/dist-packages/attrdict/dictionary.py", line 17, in __init__\n super(AttrDict, self).__init__(*args, **kwargs)\nValueError: dictionary update sequence element #0 has length 1; 2 is required\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n File "/nfvbench/nfvbench/nfvbench.py", line 611, in main\n config = config_loads(opts.config, config, whitelist_keys)\n File "/nfvbench/nfvbench/config.py", line 55, in config_loads\n raise Exception(e)\nException: dictionary update sequence element #0 has length 1; 2 is required\n'} dictionary update sequence element #0 has length 1; 2 is required The /tmp/nfvbench/nfvbench.cfg is cut and paste (OpenStack is not used and only PCIe addresses are modified) traffic_generator: generator_profile: - name: trex-local tool: TRex ip: 127.0.0.1 cores: 3 software_mode: false interfaces: - port: 0 pci: "04:00.0" - port: 1 pci: "04:00.1" intf_speed: Any help is appreciated. Regards, Charlie Li
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#24356): https://lists.opnfv.org/g/opnfv-tech-discuss/message/24356 Mute This Topic: https://lists.opnfv.org/mt/76538156/21656 Mute #nfvbench: https://lists.opnfv.org/g/OPNFV+opnfv-tech-discuss/mutehashtag/nfvbench Group Owner: opnfv-tech-discuss+ow...@lists.opnfv.org Unsubscribe: https://lists.opnfv.org/g/opnfv-tech-discuss/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
