January 2026 - openstack-discuss - lists.openstack.org

[dev][nova] Intel TDX support
by Anton Iacobaeus 18 Mar '26

18 Mar '26

Hi, I work with confidential computing and has previously contributed to Proxmox [1] to enable support for Intel TDX. I am looking to do the same for Openstack Nova. I have proposed a blueprint [2] and am drafting a spec with target for 2026.2 release. From what I gather, support for AMD SEV and SEV-ES is already implemented in Nova, while SEV-SNP support is ongoing. TDX shares some commonalities with SEV-SNP, and I believe that the work on both can be conducted in tandem. I also have experience with SEV-SNP and am happy to contribute if needed. Early feedback or any concerns and guidance are very welcome! [1] https://pve.proxmox.com/wiki/Roadmap#Proxmox_VE_9.1 [2] https://blueprints.launchpad.net/nova/+spec/intel-tdx-libvirt-support Best Regards, Anton

4 4

[Keystone][ldap] Keystone bind request to ldap not able to interpret if the user's password is expired.
by Sharath Ck 12 Mar '26

12 Mar '26

Hi, I have configured Keystone with 389ds as backend. 389ds server is configured with password policy and pwdmustchange is set to true. However on first login attempt from keystone to request token for user in 389ds is resulting in successful bind request. Perhaps, 389ds is providing control ID - control: 2.16.840.1.113730.3.4.4 in the response. But keystone is not able to interpret the control ID and provide a token with full access. Is it not possible for keystone to respond token with minimum scope to only change password? Or any response attribute to specify password is expired and needs to be changed? Kindly help with any configuration that may interpret the password expiry of ldap user and can be used in token response. P.S: configuring user_enabled_attribute to check nsAccountLock will not work as ldap user's password would be expired but not locked. user_enabled_attribute = nsAccountLock user_enabled_mask = 0 user_enabled_invert = true user_enabled_default = true Regards, Sharath

3 3

OpenStack Ops Radio Hour Survey - Timing, cadence and more
by Ildiko Vancsa 27 Feb '26

27 Feb '26

Hi All, I’m reaching out to you to draw your attention to the new OpenStack Ops Radio Hour Survey! As a reminder, the OpenStack Ops Radio Hour gathering is a meeting series in a casual setting for operators, engineers, and contributors to come together, share knowledge, and collaborate on the complexities of managing and deploying OpenStack environments. The group organizers put together a survey to better understand what channels people use to learn about upcoming meetings, what topics they would be interested in discussing, and what timing and cadence would be most suitable for folks onwards. You can find the survey here: https://openinfrafoundation.formstack.com/forms/openstack_ops_radio_hour_su… Your input is very important to make the gatherings as widely available and inclusive as possible! Please fill out the survey __by February 25, 2026.__ In the meantime, the next Ops Radio Hour is scheduled to happen on February 27th, 2026 at 1300UTC. You can find more information, agenda items and notes and recordings from previous meetings here: https://etherpad.opendev.org/p/openstack_ops_radiohour Please let me know if you have any questions or run into any challenges with the survey. Thanks and Best Regards, Ildikó

1 2

using postgresql in production
by Jiatong Shen 25 Feb '26

25 Feb '26

Hello community, We are analyzing the possibility of using PostgreSQL in production. The reason behind is that the PostgreSQL community is very open and growing fast, besides many Chinese companies provide PG compatibility. My question is what are the pros and cons between using PG and MariaDB. Are there any tools available to do online data migration for legacy environments? Any insight is deeply appreciated. Thank you. -- Best Regards, Jiatong Shen

7 8

[kolla-ansible] Migration from OVS to OVN
by Winicius Allan 16 Feb '26

16 Feb '26

Hi kolla o/ I was searching to see if Kolla currently supports migration from OVS to OVN, and I've found this email thread [1]. I'd like to know if this feature has been implemented, and if so, what the necessary steps are. [1] https://lists.openstack.org/pipermail/openstack-discuss/2023-March/032785.h… Best regards

3 3

How to access OpenStack instances from the bastion node when deploying OpenStack using Kolla-Ansible
by ussefelboukhari＠gmail.com 11 Feb '26

11 Feb '26

In test environments, some users choose to build their infrastructure inside KVM virtual machines, while others use cloud instances such as EC2 bare-metal that support nested virtualization. After deploying OpenStack, many of them face a major issue: they are unable to access their OpenStack instances. I encountered the same problem myself. After several weeks of intensive troubleshooting, I eventually found a solution. That is why I would like to share this solution with others who face the same issue, and hopefully contribute to the OpenStack community by providing recommendations that could be added to the official OpenStack documentation.

2 1

[kolla-ansible][rabbitmq] RabbitMQ cluster creation took too long
by Nguyễn Hữu Khôi 11 Feb '26

11 Feb '26

Hello. I am deploying OpenStack 2025.1 with Kolla-Ansible 2025.1 (master branch) using the latest images. However, when I created a RabbitMQ(4.0.x) cluster with 3 nodes, it took too long—around 12 minutes(it stucked at DB: virgin node -> run peer discovery). I then changed the images to RabbitMQ version 3.13.7, and it worked well. My OS: Ubuntu 24.01 Nguyen Huu Khoi

2 2

[kolla][cinder] strange behavior with 2 storage nodes
by Alexander Luedtke 05 Feb '26

05 Feb '26

Hello Everyone, I have an strange behavior here on an cinder installation deployed with Kolla: The Setup: - 1 Controller node - some compute nodes - 2 Storage nodes (storage0, storage2) - later it should be 3. both with NFS Setup what is working. both storage nodes are shown on Skyline. The Problem: 1. in Skyline both storage nodes are shown but both are shown with the same size of one of both. 2. When I created a Volume then it was created on the other host as shown (in DB, skyline and so on) Anyone an idea what I can do to correct this problems ? Thanks a lot for your help ! Alexander -- Alexander Luedtke Systemadministrator Technische Universität München TUM School of Computation, Information and Technology Chair of IT Security Garching Boltzmannstr. 3 Room 01.08.036 85748 Garching b. Muenchen Tel. +49 89 289 18039 www.sec.in.tum.de

1 1

[ironic] A first-class physical location API?
by Jay Faulkner 04 Feb '26

04 Feb '26

Something I've seen come up in a few discussions over the last couple of cycles is the need for Ironic to contain more useful information about the physical world. It's made me wonder if there's an opportunity to add bigger support for physical characteristics/locations? I'm thinking metadata like: - Datacenter the server is in - Computer room (or other DC subdivision) - Row - Rack - Slot # - Physical cabling (e.g. what switches you might be connected to and how) Do other folks see this need? I'm trying to tease out if there's something here or not, so just curious about use cases and what people think. -Jay

4 5

[neutron] IPv6 DVR for OVN - Seeking feedback on patch review prospects
by Maximilian Stinsky-Damke 04 Feb '26

04 Feb '26

Hello, I'm reaching out about the IPv6 DVR feature for OVN (bug #1998609: https://bugs.launchpad.net/neutron/+bug/1998609) and the associated patch (https://review.opendev.org/c/openstack/neutron/+/867513) As noted in the bug comments, Roberto Bartzen Acosta maintained this patch across multiple release cycles, but it hasn't received reviews from core maintainers and is currently blocked by merge conflicts. Roberto mentioned that the feature hasn't been getting traction despite the work put in. This feature would enable distributed routing for IPv6 GUA addresses in OVN deployments, similar to how IPv4 FIPs work with DVR today. The spec was approved and merged (https://review.opendev.org/c/openstack/neutron-specs/+/867979) indicating community interest. Before I invest time rebasing the patch and resolving merge conflicts, I'd like to understand: 1. Are there technical or architectural concerns that have prevented review? 2. Is there a reason this hasn't been prioritized, or is it a matter of reviewer availability? 3. Would core maintainers be willing to review if the patch is rebased and updated? I'm happy to do the rebase work, but I'd prefer not to spend time on something that won't be merged. Any guidance on the status and viability of this feature would be appreciated. Thanks, Max

2 5