| errcsool | Hi all, I seem to be hitting some issues running devstack on Ubuntu 22 virtual machine from MacOS using UTM | 00:10 |
|---|---|---|
| errcsool | Using version stable/2023.2, running stack.sh, I get an error from functions:wait_for_compute:505 "Didn't find service registered by hostname after 60 seconds" | 00:11 |
| errcsool | Then it prints and shows "nova-conductor", "nova-scheduler", and another "nova-conductor" service 'internal', 'enabled', 'up' via "$ openstack compute service list" | 00:12 |
| errcsool | Exit code 124 | 00:12 |
| errcsool | It's a default ubuntu 22 vm system, after creating a new user to run the devstack as recommended. Any help would be appreciated | 00:13 |
| clarkb | errcsool: I would look at the nova compute service logs. They may be called n-cpu on disk But you can also get them from journalctl something like journalctl -u n-cpu | 00:16 |
| clarkb | also note that ubnutu 22.04 and 22.10 are two different versions. You want 22.04 as it is an lts releas and is what we run tests on | 00:16 |
| errcsool | yes, this is ubuntu 22.04, to clarify | 00:19 |
| errcsool | ok i will look at the logs | 00:19 |
| errcsool | clarkb: two interesting messages... "WARNING nova.virt.libvirt.driver [None req-aebd87a7-ca1d-47c8-8870-2c5c205f31d3 None None] The libvirt driver is not tested on aarch64 by the OpenStack project and thus its quality can not be ensured. For more information, see: https://docs.openstack.org/nova/latest/user/support-matrix.html", followed by | 00:24 |
| errcsool | "ERROR oslo_service.service [None req-aebd87a7-ca1d-47c8-8870-2c5c205f31d3 None None] Error starting thread.: nova.exception.Invalid: The cpu_models option is not required when cpu_mode!=custom" | 00:24 |
| clarkb | the first warning is probably something you can ignore. People do run openstack on aarch64 with libvirt (there is even testing of this so the nova warning should maybe be reworded). The second thing looks like a problem to investigate | 00:25 |
| errcsool | got it | 00:25 |
| errcsool | thank you | 00:25 |
| clarkb | errcsool: it is possible that devsatck by default sets a cpu model that is for x86_64 | 00:26 |
| clarkb | you may need to unset that | 00:26 |
| errcsool | oh interesting | 00:26 |
| clarkb | ya stackrc: LIBVIRT_CPU_MODEL=${LIBVIRT_CPU_MODEL:-Nehalem} | 00:27 |
| clarkb | having a default that is consistent is necessary for things like live migration testing, but in this case I suspect is creating a problem beacuse Nehalem isn't a cpu model you want on aarch64 | 00:27 |
| errcsool | Do you know what would be the appropriate cpu model? | 00:28 |
| errcsool | ^^that makes sense | 00:28 |
| clarkb | I don't for aarch64. The first thing I would do is look in your nova config files to see what you've ended up with. custom is the default mode | 00:28 |
| clarkb | but that error implies it isn't set to custom. Maybe it is set to host-passthrough? | 00:29 |
| clarkb | ah yup I see a condition for aarch64 that sets it to host passthrough | 00:29 |
| clarkb | give me a few and I'll push up a patch you can test | 00:29 |
| errcsool | Awesome, thanks. I'm unavailable for the next hour and a half ish, but I will be sure to test it when I get back | 00:30 |
| clarkb | its also possible that you'll hit other issues on aarch64 with devsatck as I think most of the arm testing uses kolla ansible instead. But nova is the area you are most likely to have problems. Once that is workign I would expect it to be smoother sailing | 00:32 |
| opendevreview | Clark Boylan proposed openstack/devstack master: Unset nova's cpu_model config option on aarch64 https://review.opendev.org/c/openstack/devstack/+/909625 | 00:35 |
| *** haleyb is now known as haleyb|out | 00:38 | |
| errcsool | clarkb ok ill be tracking down the architecture issue | 02:28 |
| errcsool_ | clarkb: just letting you know that changing LIBVIRT_TYPE to qemu and LIBVIRT_CPU_MODE to host-passthrough in the stackrc file worked for the macos/aarch64 system running devstack. I've left LIBVIRT_CPU_MODEL=Nehalem as is | 05:17 |
| kopecmartin | gmann: i don't have rights to postpone the meeting we scheduled for today, can you reschedule it as needed? | 08:21 |
| kopecmartin | gmann: any chance you know where printing of test names together with test id is done in tempest? I have an issue with generated subunit - after I run it through subunit2junitxml the xml's name attribute contains not only the name but also other metadata e.g. [<id>,object_storage,slow] | 08:51 |
| kopecmartin | i'd need the other metadata under a different xml attribute | 08:52 |
| opendevreview | Clark Boylan proposed openstack/tempest master: Drop tempest-full-py3-opensuse15 job https://review.opendev.org/c/openstack/tempest/+/909767 | 17:03 |
| opendevreview | Clark Boylan proposed openstack/devstack master: Drop devstack-single-node-opensuse-15 nodeset https://review.opendev.org/c/openstack/devstack/+/909768 | 17:05 |
| opendevreview | Clark Boylan proposed openstack/devstack master: Drop devstack-single-node-opensuse-15 nodeset https://review.opendev.org/c/openstack/devstack/+/909768 | 17:11 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 18:15 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: POC: deploy with cpu_power_management=True https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909785 | 18:15 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: POC: deploy with cpu_power_management=True https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909785 | 20:20 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 20:20 |
| opendevreview | James Parker proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 20:27 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: POC: deploy with cpu_power_management=True https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909785 | 21:05 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 21:05 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: POC: deploy with cpu_power_management=True https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909785 | 21:08 |
| opendevreview | Artom Lifshitz proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 21:08 |
| opendevreview | James Parker proposed openstack/whitebox-tempest-plugin master: Add HyperV default enlightment tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/909225 | 21:09 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!