February 2023 - openstack-discuss

Experience with VGPUs
by Oliver Weinmann 22 Jun '23

22 Jun '23

Dear All, we are planning to have a POC on VGPUs in our Openstack cluster. Therefore I have a few questions and generally wanted to ask how well VGPUs are supported in Openstack. The docs, in particular: https://docs.openstack.org/nova/zed/admin/virtual-gpu.html explain quite well the general implementation. But I am more interested in general experience with using VGPUs in Openstack. We currently have a small YOGA cluster, planning to upgrade to Zed soon, with a couple of compute nodes. Currently our users use consumer cards like RTX 3050/3060 on their laptops and the idea would be to provide VGPUs to these users. For this I would like to make a very small POC where we first equip one compute node with an Nvidia GPU. Gladly also a few tips on which card would be a good starting point are highly appreciated. I know this heavily depends on the server hardware but this is something I can figure out later. Also do we need additional software licenses to run this? I saw this very nice presentation from CERN on VGPUs: https://indico.cern.ch/event/776411/contributions/3345183/attachments/18516… In the table they are listing Quadro vDWS licenses. I assume we need these in order to use the cards? Also do we need something like Cyborg for this or is VGPU fully implemented in Nova? Best Regards, Oliver

17 36

Kubernetes Conformance 1.24 + 1.25
by Kendall Nelson 05 Jun '23

05 Jun '23

Hello All! First of all, I want to say a huge thanks to Guilherme Steinmuller for all his help ensuring that OpenStack Magnum remains Kubernetes Certified [1]! We are certified for v1.24! However, in digging into the set of conformance tests this round, it appears that we will not be able to get 1.25 certified with the current Magnum. This means that in May when we have to re-cert (every 3 months they reach an EOL because that's k8s release cadence) we will not be able to maintain this which is the other part of why I wanted to write this email - to raise awareness and ask for help digging into what needs to happen to make sure we can retain this. Tangentially related, Guilherme has posted a patch with documentation for how to run the conformance tests for yourself and it could use some reviews [2]. And one last thing, I have begun conversation with the opendev folks about what it might take to setup jobs to run the conformance tests periodically so that we don't have to manually run them when it's time to recert. I am planning on submitting a forum topic for this for those interested :) Long live LOKI! - Kendall Nelson [1] https://github.com/cncf/k8s-conformance/pull/2446 [2] https://review.opendev.org/c/openstack/magnum/+/849156

5 10

[ironic][stable] Proposing EOL of ironic project branches older than Wallaby
by Jay Faulkner 01 Jun '23

01 Jun '23

Hi all, Ironic has a large amount of stable branches still in EM. We need to take action to ensure those branches are either retired or have CI repaired to the point of being usable. Specifically, I'm looking at these branches across all Ironic projects: - stable/queens - stable/rocky - stable/stein - stable/train - stable/ussuri - stable/victoria In lieu of any volunteers to maintain the CI, my recommendation for all the branches listed above is that they be marked EOL. If someone wants to volunteer to maintain CI for those branches, they can propose one of the below paths be taken instead: 1 - Someone volunteers to maintain these branches, and also report the status of CI of these older branches periodically on the Ironic whiteboard and in Ironic meetings. If you feel strongly that one of these branches needs to continue to be in service; volunteering in this way is how to save them. 2 - We seriously reduce CI. Basically removing all tempest tests to ensure that CI remains reliable and able to merge emergency or security fixes when needed. In some cases; this still requires CI fixes as some older inspector branches are failing *installing packages* in unit tests. I would still like, in this case, that someone volunteers to ensure the minimalist CI remains happy. My intention is to let this message serve as notice and a waiting period; and if I've not heard any response here or in Monday's Ironic meeting (in 6 days), I will begin taking action on retiring these branches. This is simply a start; other branches (including bugfix branches) are also in bad shape in CI, but getting these retired will significantly reduce the surface area of projects and branches to evaluate. I know it's painful to drop support for these branches; but we've provided good EM support for these branches for a long time and by pruning them away, we'll be able to save time to dedicate to other items. Thanks, Jay Faulkner

3 5

[magnum] Propose deprecation of Docker Swarm, CoreOS, fedora-ironic
by Jake Yip 28 Apr '23

28 Apr '23

Hi, Due to limited resources, Magnum team has decided to put most effort into supporting the most popular use case - Kubernetes on Fedora CoreOS. In light of that, we are seeking maintainers for the less well maintained drivers / COE - Docker Swarm, CoreOS and fedora-ironic drivers. The reasons are as follows: - Docker Swarm - It uses Fedora Atomic which has been EOL [1] - CoreOS - It is EOL [2] - fedora-ironic - Not tested, unsure if it is still working If anyone wants to take up maintenance of any of the above, we will be glad. Please let us know either through the ML or at the Magnum meetings. If there are no suitable maintainers, we will start the deprecation of these before end of Antelope. Regards, Jake [1]: https://projectatomic.io/blog/2019/11/fedora-atomic-host-nearing-eol/ [2]: https://www.redhat.com/en/technologies/cloud-computing/openshift/what-was-c… -- Jake Yip Magnum Team

2 3

Re: [all][stable][ptl] Propose to EOL Rocky series
by Elõd Illés 17 Apr '23

17 Apr '23

Hi, thanks for all the feedbacks from teams so far! @Zigo: Extended Maintenance process was created just for the same situation: to give space to interested parties to cooperate and keep things maintained even when stable releases are over their 'supported' lifetime. So it's good to see that there is interest in it! Unfortunately, with very old branches we've reached the state where gates can't be maintained and without a functional gate it's not safe to merge patches (yes, even security fixes) and they are just using resources (CI & maintainers' time). When gate is broken in such extent, then i think the community have to accept that it is not possible to merge patches confidently and needs to EOL that release. (Yes, it means that some vendors still have to do it for their own sake, testing it by their own, etc). Keeping things open in such state gives the false signal to vendors that 'it is maintained' however this is done without fully testing the changes. Another aspect is that code cannot be cleaned up until those old branches are still present (CI jobs, project configurations, etc) which gives pain for developers. So, however some vendors would appreciate probably to keep things open forever, for the community this is not beneficial and doable I think. Seeing that teams want to EOL their old branches is a proof that this is needed (unfortunately). Thanks, Előd Illés irc: elodilles @ #openstack-stable / #openstack-release ________________________________ From: Thomas Goirand <thomas(a)goirand.fr> Sent: Saturday, January 28, 2023 2:25 PM Cc: openstack-discuss(a)lists.openstack.org <openstack-discuss(a)lists.openstack.org> Subject: Re: [all][stable][ptl] Propose to EOL Rocky series Hi, I understand the gate is broken, however, Rocky is in Debian LTS, and I would like to keep the possibility to merge patches, even with the gate tests disabled. CVE-2022-47951 is an example why this is important... Thomas Goirand (zigo) On Jan 28, 2023 11:54, Radosław Piliszek <radoslaw.piliszek(a)gmail.com> wrote: > > Masakari is happy to EOL Rocky too. > > Radek > -yoctozepto > > On Fri, 27 Jan 2023 at 20:59, Jay Faulkner <jay(a)gr-oss.io> wrote: > > > > Thanks for doing all this cleanup work Elod. Ironic is OK with retirements of these shared resources up to Train. > > > > - > > Jay Faulkner > > > > On Fri, Jan 27, 2023 at 10:12 AM Elõd Illés <elod.illes(a)est.tech> wrote: > >> > >> Hi, > >> > >> Similarly like the Queens branch EOL proposal [1] I would like to propose > >> to transition every project's stable/rocky to End of Life: > >> > >> - gates are mostly broken > >> - minimal number of activity can be seen on this branch > >> - some core projects already transitioned their stable/rocky to EOL > >> recently (like ironic, neutron, nova) > >> - gate job definitions are still using the old, legacy zuul syntax > >> - gate jobs are based on Ubuntu Xenial, which is also beyond its public > >> maintenance window date and hard to maintain > >> > >> Based on the above, if there won't be any project who wants to keep open > >> their stable/rocky, then I'll start the process of EOL'ing Rocky stable > >> series as a whole. If anyone has any objection then please respond to > >> this mail. > >> > >> Thanks, > >> > >> Előd Illés > >> irc: elodilles @ #openstack-stable / #openstack-release > >> > >> [1] https://lists.openstack.org/pipermail/openstack-discuss/2022-October/031030… > >> >

3 4

DCN compute service goes down when a instance is scheduled to launch | wallaby | tripleo
by Swogat Pradhan 27 Mar '23

27 Mar '23

Hi, I currently have 3 compute nodes on edge site1 where i am trying to launch vm's. When the VM is in spawning state the node goes down (openstack compute service list), the node comes backup when i restart the nova compute service but then the launch of the vm fails. nova-compute.log 2023-02-26 08:15:51.808 7 INFO nova.compute.manager [req-bc0f5f2e-53fc-4dae-b1da-82f1f972d617 - - - - -] Running instance usage audit for host dcn01-hci-0.bdxworld.com from 2023-02-26 07:00:00 to 2023-02-26 08:00:00. 0 instances. 2023-02-26 08:49:52.813 7 INFO nova.compute.claims [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] [instance: 0c62c1ef-9010-417d-a05f-4db77e901600] Claim successful on node dcn01-hci-0.bdxworld.com 2023-02-26 08:49:54.225 7 INFO nova.virt.libvirt.driver [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] [instance: 0c62c1ef-9010-417d-a05f-4db77e901600] Ignoring supplied device name: /dev/vda. Libvirt can't honour user-supplied dev names 2023-02-26 08:49:54.398 7 INFO nova.virt.block_device [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] [instance: 0c62c1ef-9010-417d-a05f-4db77e901600] Booting with volume c4bd7885-5973-4860-bbe6-7a2f726baeee at /dev/vda 2023-02-26 08:49:55.216 7 WARNING nova.cache_utils [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] Cache enabled with backend dogpile.cache.null. 2023-02-26 08:49:55.283 7 INFO oslo.privsep.daemon [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] Running privsep helper: ['sudo', 'nova-rootwrap', '/etc/nova/rootwrap.conf', 'privsep-helper', '--config-file', '/etc/nova/nova.conf', '--config-file', '/etc/nova/nova-compute.conf', '--privsep_context', 'os_brick.privileged.default', '--privsep_sock_path', '/tmp/tmpin40tah6/privsep.sock'] 2023-02-26 08:49:55.791 7 INFO oslo.privsep.daemon [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] Spawned new privsep daemon via rootwrap 2023-02-26 08:49:55.717 2647 INFO oslo.privsep.daemon [-] privsep daemon starting 2023-02-26 08:49:55.722 2647 INFO oslo.privsep.daemon [-] privsep process running with uid/gid: 0/0 2023-02-26 08:49:55.726 2647 INFO oslo.privsep.daemon [-] privsep process running with capabilities (eff/prm/inh): CAP_SYS_ADMIN/CAP_SYS_ADMIN/none 2023-02-26 08:49:55.726 2647 INFO oslo.privsep.daemon [-] privsep daemon running as pid 2647 2023-02-26 08:49:55.956 7 WARNING os_brick.initiator.connectors.nvmeof [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] Process execution error in _get_host_uuid: Unexpected error while running command. Command: blkid overlay -s UUID -o value Exit code: 2 Stdout: '' Stderr: '': oslo_concurrency.processutils.ProcessExecutionError: Unexpected error while running command. 2023-02-26 08:49:58.247 7 INFO nova.virt.libvirt.driver [req-3a1547ea-326f-4dd0-9127-7f4a4bdf1e45 b240e3e89d99489284cd731e75f2a5db 4160ce999a31485fa643aed0936dfef0 - default default] [instance: 0c62c1ef-9010-417d-a05f-4db77e901600] Creating image Is there a way to solve this issue? With regards, Swogat Pradhan

5 40

[all][ptg] Pre-PTG discussion: New Keystone Middleware Feature Supporting OAuth2.0 with External Authorization Service
by Hiromu Asahina 27 Mar '23

27 Mar '23

Hello Everyone, Recently, Tacker and Keystone have been working together on a new Keystone Middleware that can work with external authentication services, such as Keycloak. The code has already been submitted [1], but we want to make this middleware a generic plugin that works with as many OpenStack services as possible. To that end, we would like to hear from other projects with similar use cases (especially Ironic and Barbican, which run as standalone services). We will make a time slot to discuss this topic at the next vPTG. Please contact me if you are interested and available to participate. [1] https://review.opendev.org/c/openstack/keystonemiddleware/+/868734 -- Hiromu Asahina

4 11

[TripleO] Last maintained release of TripleO is Wallaby
by James Slagle 24 Mar '23

24 Mar '23

The team of TripleO developers, reviewers, and community have no plans to do another release of TripleO, or to continue maintaining the Zed release. This means the last maintained release of TripleO is Wallaby. The plan is to continue to maintain the train and wallaby branches of TripleO repositories under the openstack namespace. We plan to continue to run existing CI on those branches, although we anticipate reducing to a minimal set of jobs in the future. Going forward, the TripleO team expects to continue involvement in other OpenStack projects, and focus on different approaches to OpenStack deployment. The team is not aware of enough other community interest to continue to maintain the zed, master, and future branches of TripleO. Eventually, we will empty commit these branches with a pointer to Wallaby being the last release. If other interests do exist to maintain TripleO going forward, then those plans can adjust. TripleO does not plan on nominating a PTL in the current Bobcat election cycle. If a PTL-like contact is needed for train and wallaby maintenance, then someone will be nominated. I am also not sure how to reflect this project structure change within openstack/governance, and am looking for guidance on how to do so. This type of change did not seem to exactly fit within the concepts of inactive or retired projects. I can work on the patches for governance with some direction, if anyone can provide it. Thanks, and please let us know any questions or feedback about this plan. -- -- James Slagle --

6 21

[Openstack] Lack of Balance solution such as Watcher.
by Nguyễn Hữu Khôi 16 Mar '23

16 Mar '23

I am building a private cloud.Everything is ok. But I cannot find a way to balance vm when It have heavy load which cause impact other vm on the compute node. I found solutions such as Watcher and Leveller. But they need to be done manually. Watcher is not good because It need cpu metric such as cpu load in Ceilometer which is removed so we cannot use it. Leveller is good, but It is obsolete. Nguyen Huu Khoi

4 7

live resize
by Parsa Aminian 15 Mar '23

15 Mar '23

hello on openstack with ceph backend is it possible to live resize instances ? I want to change flavor without any down time .

5 6