| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-05-06 | |||
| 15:17:47 | openstackgerrit | Jiri Suchomel proposed openstack/nova master: Add ability to download Glance images into the libvirt image cache via RBD https://review.opendev.org/574301 | |
| 16:44:26 | stephenfin | bauzas: Looks like the multiple vGPU tests are a bit racey https://zuul.opendev.org/t/openstack/build/48123952eb1048efb76be65bb1893c99/console | |
| 16:44:30 | stephenfin | *functional tests | |
| 16:44:50 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/ussuri: Remove stale nested backport from InstancePCIRequests https://review.opendev.org/725931 | |
| 16:44:59 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/train: Remove stale nested backport from InstancePCIRequests https://review.opendev.org/725932 | |
| 16:45:48 | openstackgerrit | Stephen Finucane proposed openstack/nova stable/train: Remove stale nested backport from InstancePCIRequests https://review.opendev.org/725932 | |
| 16:59:35 | bauzas | stephenfin: ack, I'm off the keyboard now, but please revert the test_multiple_instance_create() method if you think it's racey a lot | |
| 17:18:26 | openstackgerrit | Takashi Kajinami proposed openstack/nova master: conf: Add three new '[libvirt] live_migration_*' options https://review.opendev.org/456571 | |
| 18:03:30 | openstackgerrit | Takashi Kajinami proposed openstack/nova master: conf: Add four new '[libvirt] live_migration_*' options https://review.opendev.org/456571 | |
| 21:50:47 | zigo | artom: https://review.opendev.org/724684 +1 ? :) | |
| 22:12:14 | artom | zigo, I guess? What did we agree to about docs? | |
| 22:12:24 | artom | IIUC we're doing to document it in the project docs for now? | |
| 22:12:44 | zigo | I don't remember TBH. | |
| 22:13:16 | zigo | I was really busy yesterday having Keystone's latest security fix tested in Debian ... (now I can publicly talk about it...) | |
| 22:14:07 | zigo | artom: What should I document exactly? That /healthcheck can be used? | |
| 22:14:13 | zigo | How to disable it? | |
| 22:15:48 | artom | zigo, so I would say release note at the very least (are you familiar with our reno tool?) | |
| 22:15:57 | artom | zigo, and then something like https://docs.openstack.org/keystone/latest/admin/health-check-middleware.html | |
| 22:16:29 | zigo | I did write some notes with reno in the past, though that skill is probably rusted... | |
| 22:16:31 | zigo | :) | |
| 22:17:05 | artom | zigo, also, I'd be weary of making you do more work before there's core consensus | |
| 22:17:20 | artom | I'm just a random dev, my opinion holds less weight ;) | |
| 22:18:15 | artom | OTOH, the attempt to reach consensus appears to be older than my daughter, so... :) | |
| 22:20:35 | zigo | artom: How old is she? :) | |
| 22:20:44 | artom | zigo, 6 | |
| 22:20:48 | zigo | Like mine... | |
| 22:21:04 | zigo | Got an older boy too (he's 8). | |
| 22:21:22 | artom | Our is 4 | |
| 22:22:46 | openstackgerrit | Thomas Goirand proposed openstack/nova master: Add a /healthcheck URL https://review.opendev.org/724684 | |
| 22:22:49 | zigo | Release note added. | |
| 22:23:15 | zigo | I also did the same kind of work for Cinder and wrote a functional test there too! This time, on my own... :P | |
| 22:23:54 | artom | \o/ | |
| 22:23:58 | artom | Transferable skills! | |
| 22:37:47 | artom | zigo, replied, see review. Sorry no +1 yet :( And now, dinner :) | |
| 22:41:30 | zigo | ok | |
| #openstack-nova - 2020-05-07 | |||
| 03:41:53 | openstackgerrit | tianmaofu proposed openstack/nova master: Delete queued image after deleting instance when creating image from instance https://review.opendev.org/724839 | |
| 05:11:14 | HagunKim | Hello Nova, I have a question. When I set my flavor vcpu=19, the vm only has 1 cpu core in linux. It is okay vcpu 18 or 17 flavor. What wrong with my Openstack? | |
| 05:15:29 | HagunKim | libvirt vm info has 19 vpus. | |
| 07:38:16 | bauzas | good morning Nova | |
| 07:40:16 | bauzas | gibi: fwiw, stephenfin found some races for the multiple_instance_create() vGPU functest http://logstash.openstack.org/#dashboard/file/logstash.json?query=message%3A%5C%22testtools.matchers._impl.MismatchError%3A%202%20!%3D%201%5C%22%20AND%20build_name%3A%5C%22nova-tox-functional-py36%5C%22 | |
| 07:40:25 | bauzas | not a lot tho (8 for 7 days) | |
| 07:50:19 | gibi | bauzas: good morning. I will check that race soon | |
| 08:09:53 | bauzas | gibi: nah, don't worry, I'll look | |
| 08:10:11 | bauzas | if I can't find a solution soon, I'll write an e-r change | |
| 08:11:29 | gibi | OK | |
| 08:22:03 | gibi | bauzas: I was able to recreate the race locally | |
| 08:23:11 | bauzas | gibi: I wonder, when i tested, I wasn't having this race | |
| 08:25:16 | gibi | could be something else changed around the test. In the recent weeks I did not run a lot of full functional suites so I don't know how far this problem goes back | |
| 08:27:16 | bauzas | gibi: when you reproduced, have you run only this test, or more ? | |
| 08:27:37 | gibi | I run the whole functional suite | |
| 08:27:47 | gibi | let me try running the test alon repeatedly | |
| 08:28:01 | bauzas | ok, because I only run the two multiple_create() methods | |
| 08:28:12 | bauzas | (looping over) | |
| 08:28:33 | bauzas | so, I wonder whether some global object would be raced | |
| 08:29:24 | gibi | yeah, that is possible | |
| 08:29:46 | gibi | I have to jump on a call now, but I will run the tests in different ways to see if it is an interference or not | |
| 08:30:12 | bauzas | kk thanks | |
| 08:33:17 | openstackgerrit | Jie Li proposed openstack/nova master: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/722187 | |
| 08:47:32 | bauzas | ah, bingo https://docs.openstack.org/os-testr/latest/user/subunit_trace.html | |
| 08:49:39 | bauzas | or rather https://stestr.readthedocs.io/en/latest/MANUAL.html#running-previously-failed-tests | |
| 08:50:03 | gibi | bauzas: running just the test_multiple_instance_create* test case I can still reproduce the problem | |
| 08:50:25 | gibi | but it needed like ~50 execution to get it once | |
| 08:51:24 | gibi | now trying just only the single test_multiple_instance_create$ case as I suspect that it is not an interference | |
| 08:52:19 | gibi | when I run the two test cases the, the test_multiple_instance_create$ failed, and the two case run in two separate executor so they cannot interfere | |
| 09:03:40 | bauzas | gibi: you can ask to use a single subunit | |
| 09:04:12 | gibi | bauzas: just paste me the command you want me to run and I will try that as well | |
| 09:04:12 | bauzas | gibi: that's why I'd like to get the subunit output from the failing job and trying to rerun it | |
| 09:04:47 | bauzas | gibi: https://stestr.readthedocs.io/en/latest/MANUAL.html#parallel-testing | |
| 09:04:48 | gibi | so far, after 77 execution no failure in the single test_multiple_instance_create$ test case execution | |
| 09:05:53 | bauzas | gibi: I got one after a shit number of executions | |
| 09:08:59 | Yumeng | Hi gibi! Good morning, I'm Yumeng from Cyborg Team. | |
| 09:09:20 | Yumeng | Cyborg wants to have a "cyborg/neutron/nova sriov integration" discussion with NOVA and NEUTRON during the Virtual PTG. | |
| 09:09:24 | Yumeng | Could you please find 1 hour for nova team to discuss this topic with cyborg and neutron team? | |
| 09:11:33 | Yumeng | I notice that there is a nova-neutron session on PTG Friday 13:00 UTC, can we join after that on Friday 15:00 UTC? | |
| 09:12:08 | gibi | Yumeng: hi! the schedule already closed so we have to use an exsiting booked slot for that discussion. Do you have a room booking from Friday 15:00 ? | |
| 09:13:42 | Yumeng | gibi: no, I don't have now. But I can email Kendall and book the room if it's ok for nova and neutron. | |
| 09:15:48 | gibi | Yumeng: I think we can try Friday 14:00 - 15:00 where we already have a room booked, or if you can get a room from Kendall then I we can do the discussion from 15:00 - 16:00 | |
| 09:17:19 | Yumeng | gibi: aha! Friday 14:00 - 15:00 is fantastic! I am just not sure if it is already occupied. | |
| 09:18:07 | gibi | Yumeng: I kept that as an overflow slot so I think we can use that for cyborg. If you need the neutron team present as well then you have to synch with slaweq too | |
| 09:19:21 | bauzas | Yumeng: tbh, what use case are you talking of ? | |
| 09:19:24 | Yumeng | Great! Thank you so much gibi! I will sync with slaweq! ^^ | |
| 09:19:33 | gibi | Yumeng: cool. I added it to the nova etherpad | |
| 09:19:41 | bauzas | SR-IOV but how ? | |
| 09:19:52 | Yumeng | bauzas: yes, exactly! | |
| 09:20:03 | gibi | Yumeng, bauzas: yeah that is my next request, please try to prepare some material before the discussion | |
| 09:20:19 | bauzas | Yumeng: okay, but why do you need all contributors ? | |
| 09:20:24 | bauzas | in nova and neutron ? | |
| 09:20:42 | bauzas | what kind of feature request would you want to do in Cyborg ? | |
| 09:20:53 | gibi | ^^ +1 | |
| 09:21:02 | Yumeng | We will try to write a spec to describe this topic. And one of our team member from Intel is preparing a POC. | |
| 09:21:45 | Yumeng | yes, a lot question should be answered. ^^ | |
| 09:22:34 | bauzas | Yumeng: before writing a spec, can you just summarize the reason ? | |
| 09:28:46 | bauzas | Yumeng: just saying, we still need to work on providing the NUMA topologies in Placement and then providing the PCI devices | |
| 09:29:25 | bauzas | and when we discussed about SR-IOV to be in Placement, we said 'well, why not, but it's a mid-term issue' | |
| 09:29:37 | bauzas | at least we first need to do the above before ^ | |
| 09:29:55 | bauzas | so, even if we would like to do it for Victoria, it wouldn't be possible | |
| 09:30:30 | bauzas | Yumeng: that's why I just want to be honest to say that I'm not sure we could do anything for you during Victoria | |
| 09:30:47 | bauzas | and then, not sure we would need to ask all the contributors to be around just for this | |
| 09:31:03 | bauzas | sorry, 'this' being the virtual PTG | |
| 09:31:57 | bauzas | Yumeng: what we *could* do tho is to discuss about that once you provide some docs, maybe just an etherpad for explaining what you'd like | |