September 2024 - openstack-discuss

[manila] Dalmatian bugsquash - Sep 9th - 10th
by Carlos Silva 05 Sep '24

05 Sep '24

Hello, Zorillas and interested stackers! We'll have the second bugsquash of the Dalmatian cycle next week. As we have agreed a couple of weeks ago, it will be a two day event (Sep 9 and 10th). We will have a kick-off meeting on Monday, 15 UTC and we will use a meetpad [1] for the meeting. We will be focusing on bugs that are in progress and try to get them delivered as part of the 2024.2 Dalmatian release. Please check the ethercalc with the list of bugs [2] and add the bug you are willing to work on over the next week if it is not there yet. I'm looking forward to it :) [1] https://meetpad.opendev.org/manila-dalmatian-m3-bugsquash [2] https://ethercalc.net/mgstvday5vq0 Regards, carloss

1 0

FIPs broken when used with tunneled router gateway networks
by Sesterhenn, Maximilian 05 Sep '24

05 Sep '24

Hey all, I am currently facing an issue with OpenStack Neutron and OVN and could need some help here. Maybe someone else is doing something similar and can report if it works for them. TLDR: FIPs are broken for us when used together with routers which use a tunneled network as router gateway network. ovn_router_indirect_snat is not used and both OVN 23.09 and 24.03 have been tested. FIPs work fine when used with an external provider network. We have deployed three OpenStack projects, let's just call the left, middle and right for simplicity. The middle project has a network with a large subnet. No special configuration here, just a big subnet which is transported by OVN through Geneve. The idea is to use this subnet as an internal transfer network between different projects / tenants, which means we connect routers with SNAT and FIP configuration. Left and right are potentially the same subnet, that's why we cannot simply route the traffic. The network is shared with "access_as_external/shared" RBAC rules with both the left and right tenant, so it can be selected as the external gateway network for routers. I'm aware that until very recently, neutron was unable to assign a chassis to routers which use gateway networks which are tunneled. Thats why we are currently running the latest neutron code from the master branch where this feature was added. Let's say we want to connect from left to right and we have a router in each project that has this large network from the middle project as the external gateway network. An instance in the right tenant has a FIP to be reachable. Another instance in the left tenant is masqueraded by the SNAT functionality of the router. The router in the left project performs SNAT and the router in the right project performs the FIP DNAT. While the instance from the left project can ping the internal and external interfaces of its local router, the FIP is not reachable. The external interface of the remote router (from the right project) can be pinged as well. When capturing packets on the underlying hosts we can see the packet going from the compute host of the left instance to one of the network nodes. SNAT happens there, DNAT is not visible, this could be due to the packet capture scenario. However, we see ARP requests coming from the external interface of the router on the right project for the FIP address. We see the same ARP requests on a second network node. To me it would make more sense if the router from the left project would ARP for the FIP address. This ARP request is not being answered. We're using enable_distributed_floating_ip / DVR, so the FIP is actually configured onto the compute node. We're not using ovn_router_indirect_snat, here are issues with that. We even tried that; it makes no difference. OVN 23.09 (tested 24.03 as well) OpenStack 2024.1 (neutron master) deployed using kolla-ansible. BR Maximilian

2 2

[charms] Proposing Brian Haley and Nicolas Bock for OpenStack Charms core
by Edward Hope-Morley 05 Sep '24

05 Sep '24

Hi everyone, I would like to propose Brian Haley and Nicolas Bock for the OpenStack Charms core development team. Both Brian and Nicolas have been members of the stable charms core team since 2023 and been hugely helpful in that role, providing insightful feedback to reviews as well as contributions of their own. Nicolas has also been a contributor to the OpenStack charms for many years and Brian is well versed in the ways of upstream OpenStack maintenance through his role as Neutron PTL so I have full confidence that they will prove valuable additions to the charms-core team. Thank you both for your contributions to the project! Regards, Ed

3 2

[Kolla] [LetsEncrypt] Internal/External/Backend Certificates
by jjjamesg＠proton.me 05 Sep '24

05 Sep '24

Hi all, another one here that can't seem to get kolla running with letsencrypt certs, just curious as to what settings are required for kolla to pull internal/external/backend certs from letsencrypt? I've looked at the docs and have tried a few settings but keep getting errors, sure i had this working last week before i tore it down ready to deploy for production and now i just cant get it to complete the deployment because of certs, only thing i can think of is i need to run kolla-ansible -i multinode certificates but i thought that was just for self made certs? Also, what are best practices for backend tls certs, self signed or using the internal cert etc?? Is it possible to use letsencrypt at all for backend? Cheers all

3 6

[neutron] OpenStack OVS DVR setup in large deployments
by frantisek.reznicek.szn＠gmail.com 05 Sep '24

05 Sep '24

Dear all, we are running several OpenStack instances in Czechia. New generation of clouds we launched are deployed OpenStack Yoga on top of k8s via openstack-helm, using Neutron OVS DVR. Very recently we noticed "wasting" public IP addresses for "network:floatingip_agent_gateway ports" as briefly discussed earlier here [1] while we were scaling number of nodes up (currently around 40, final state around 300). The linked post[1] suggests to use Service subnets and let allocate "network:floatingip_agent_gateway ports" from internal address range. The service subnets documentation page[2] comes with example where "network:floatingip_agent_gateway ports" being allocated from bogon ip range 198.51.100.0/24. I want to achieve state where no allow allocation of "network:floatingip_agent_gateway ports" is allocated from public IP range. I'd preferably stay using OVS+DVR if possible. 1] Could you possibly confirm that DVR setup will accept "network:floatingip_agent_gateway ports" being allocated from smaller network in private ipv4 class A range 10.0.0.0/8? At the moment "network:floatingip_agent_gateway ports" are the ports where all FIP trafic goes, so it is kind of surprising that private addresses can be used there. Could you elaborate more in detail on that? 2] We also use multiple physical provider network segments within one logical openstack network. Can you comment on whether Service subnets feature and multi-segment feature are compatible a can be used together? I believe service subnet for "network:floatingip_agent_gateway ports" has to be routed as segment too, is that right? 3] What steps I would need to do to start using OVS w/o DVR apart from disabling distributed routing in neutron and in neutron-ovs-agent and in L3 agent and restart? Is there any special procedure? Thanks upfront for your comments... Kind Regards, František [1] https://lists.openstack.org/pipermail/openstack-dev/2016-June/096384.html [2] https://docs.openstack.org/neutron/yoga/admin/config-service-subnets.html#e…

2 2

[cinder] Feature Freeze Exception Request
by Zong, Yian 05 Sep '24

05 Sep '24

Dear All, I'd like to request a feature freeze exception for dalmatian-3 milestone regarding: 907133: Dell PowerStore driver: Add Cinder active-active support | https://review.opendev.org/c/openstack/cinder/+/907133 This change adds cinder active-active support for Dell PowerStore driver and allows users to configure PowerStore backends in cinder clustered environment. Please let me know if you have any concern. Thank you! BR, Yian Internal Use - Confidential

1 0

[all][elections][ptl][tc] Conbined PTL/TC 2025.1 cycle Election Voting Kickoff
by Slawomir Kaplonski 05 Sep '24

05 Sep '24

Poll for TC elections is now open and will remain open for you to cast your vote until Sep 18, 2024 23:45 UTC. We are selecting 4 TC members. Please rank all candidates in your order of preference. You are eligible to vote in the TC election if you are a Foundation individual member[0] that also has committed to any official project team's deliverable repositories[1] over the Sep 15, 2023 00:00 UTC - Aug 28, 2024 00:00 UTC timeframe (2024.1 to 2024.2) or if you are in the list of extra-acs[2] for any official project team. If you are eligible to vote in an election, you should find your email with a link to the Condorcet Internet Voting Service (CIVS) page to cast your vote in the inbox of your Gerrit preferred email[3]. Emails should be already sent to everyone on Aug 30. If you have already voted using the CIVS link from that email, you don't need to submit your vote again now. What to do if you don't see the email and have a commit in at least one of the projects having an election: * due to a new CIVS policy, to get the email from poll supervisors, you must opt in to email communication opt in with your Gerrit email address at: https://civs1.civs.us/cgi-bin/opt_in.pl * eligible voters who opt in to email communication late can see a pending poll invitation and vote until Sep 18, 2024 23:45 UTC. * check the trash or spam folders of your Gerrit Preferred Email address, in case it went into trash or spam * wait a bit and check again, in case your email server is a bit slow * find the sha of at least one commit from the project's deliverable repos[0] and email the election officials[4]. If we can confirm that you are entitled to vote, we will add you to the voters list for the appropriate election. Our democratic process is important to the health of OpenStack, please exercise your right to vote! Candidate statements/platforms can be found linked to Candidate names on this page: https://governance.openstack.org/election/ Happy voting, [0] https://www.openstack.org/community/members/ [1] The list of the repositories eligible for electoral status: https://opendev.org/openstack/governance/src/tag/to-be-released/reference/p… [2] Look for the extra-acs element in [1] [3] Sign into review.openstack.org: Go to Settings > Contact Information. Look at the email listed as your preferred email. That is where the ballot has been sent. [4] https://governance.openstack.org/election/#election-officials -- Best regards Slawek Kaplonski skaplons(a)redhat.com

1 0

[cinder] Offering local storage using the device mapper clone target
by Jan Horstmann 04 Sep '24

04 Sep '24

Hello cinder team and other interested parties, I have recently started thinking about ways to offer local storage as part of the OpenStack block storage service. Naturally any solution would require to transparently migrate volumes to and with the instances they are attached to, so that the user experience of being able to access their storage from anywhere would not be degraded. Fortunately, with the device mapper clone target ([1]), a mechanism is available in the linux kernel which allows to do exactly that: Create copies of devices (even network attached ones) while allowing to immediately use the target device. I have detailed the motivation and some implications of having local storage in the block storage service in a blog post [2], which I do not want to repeat here in order to not make this mail longer than it needs to be. The blog post also outlines a way to implement a cinder driver leveraging this mechanism, which is loosely based on the way cinder migrates volumes between backends, i.e.: When a volume is going to be attached, a new volume is created on the instance's host, which is backed by a device mapper clone target with the original volume as its source; Then the volume objects are switched and a local connector is returned, which may be used immediately. Since then I went ahead and created a proof of concept cinder driver ([3]), which roughly follows what is outlined in the post (some lessons were already learned during implementation). The driver is based on the in-tree lvm driver for basic volume management and can do basic things like create/delete, attach/detach and also handles live migration and rebuilding of instances. There is nothing currently regarding snapshots and volume migrations will definitly not work as expected yet, but it should be sufficient to get a feeling for the general idea. I am currently looking for feedback on the idea and the implementation and would like to know if this is of interest to others. I am happpy to answer any questions or present this in more detail. In case anyone wants to take the driver for a spin I am going to leave some instructions on how to set up the environment in [4]. Regards, Jan [1] https://docs.kernel.org/admin-guide/device-mapper/dm-clone.html [2] https://scs.community/2024/07/12/local-block-storage-via-dm-clone/ [3] https://github.com/janhorstmann/openstack-cinder/tree/59e1808a24d112d2e8d08… [4] - Set up an OpenStack environment - Create an lvm2 volume group and run a cinder volume service on every hypervisor - Configure a new backend on every volume service with the same `volume_backend_name` Configure it like you would an lvm backend, but with `volume_driver = cinder.volume.drivers.dmclone.DMCloneVolumeDriver`, e.g.: [local-volumes] volume_driver = cinder.volume.drivers.dmclone.DMCloneVolumeDriver volume_group = local volume_backend_name = local-volumes target_helper = tgtadm target_protocol = iscsi lvm_type = default I have tested with the values above, but any target driver should work -- Jan Horstmann Senior Cloud Engineer Mail: horstmann(a)osism.tech Web: https://osism.tech OSISM GmbH Talweg 8 / 75417 Mühlacker / Deutschland Geschäftsführer: Christian Berendt Unternehmenssitz: Mühlacker Amtsgericht Mannheim, HRB 750852

1 1

[OSSA-2024-003] OpenStack Ironic: Unvalidated image data passed to qemu-img (CVE-2024-44082)
by Brian Rosmaita 04 Sep '24

04 Sep '24

======================================================== OSSA-2024-003: Unvalidated image data passed to qemu-img ======================================================== :Date: September 04, 2024 :CVE: CVE-2024-44082 Affects ~~~~~~~ - Ironic: <21.4.3, >=22.0.0 <23.0.2, >=23.1.0 <24.1.2, >=25.0.0 <26.0.1 - Ironic-python-agent: <9.4.2, >=9.5.0 <9.7.1, >=9.8.0 <9.11.1, >=9.12.0 <9.13.1 Description ~~~~~~~~~~~ Dan Smith and Julia Kreger of Red Hat and Jay Faulkner of G-Research noticed a vulnerability in image processing for Ironic, in which a specially crafted image could be used by an authenticated user to exploit undesired behaviors in qemu-img, including possible unauthorized access to potentially sensitive data. Mitigation ---------- The attached patches contain code allowing Ironic and the Ironic-Python-Agent (IPA) to pre-screen images before passing them to qemu-img. You should patch both of them because in some popular deployment configurations, it is possible for images to bypass the Ironic conductor for more efficient image downloads. In situations where it is not possible to patch IPA, there is a new configuration option, ``[conductor]conductor_always_validates_images``. When this option is True, all image downloads are forced to occur through the Ironic conductor, where they are validated. However, this may incur a significant performance hit; hence our recommendation to patch both Ironic and IPA. Patches are available for both Ironic and IPA for all maintained branches, namely, the Dalmatian development branch (current master) through Antelope. For these branches, ``conductor_always_validates_images`` defaults to False. For the unmaintained branches (Zed through Victoria), patches are not available for IPA due to a significant risk of regressions. For these branches, the Ironic patches set ``conductor_always_validates_images`` to True, as this is the only safe way to run Ironic with an un-patched IPA. Cached images ------------- Images in the Ironic image cache should be purged. To purge the cache, stop the Ironic conductor and remove the files in the ``[pxe]/instance_master_path`` directory on that conductor node. Supported image formats ----------------------- A new configuration option, ``[conductor]permitted_image_formats`` controls what image formats will be accepted by Ironic. By default, the allowed formats are 'raw' and 'qcow2', which are the only formats that Ironic is tested with. In previous releases, it was possible to use images of other, unsupported formats with Ironic, but this is now blocked by default due to security issues arising from the image conversion process. It is possible, though not recommended, to expand the list of permitted formats. Alternatively, users requiring the use of an image in an unsupported format can convert the image offline into a supported format before uploading the image to Glance or otherwise making the image available to Ironic. Consult the Ironic documentation for details. Warning ------- As is well-documented, the OpenStack Ironic project does not support the use of ironic-lib for non-Ironic use cases. In its normal supported context, ironic-lib assumes that images have been pre-screened before ironic-lib comes in contact with them. Thus, unsupported use of ironic-lib leaves you vulnerable to the exploit outlined in this OSSA. The Ironic project intends eventually to explicitly remove the vulnerable methods in ironic-lib, but reminds the reader that independent use of ironic-lib is not supported. Bugfix branches --------------- Patches are available for maintained bugfix branches. These patches will be merged to the maintained branches, but no new releases will be triggered from those branches. Patches ~~~~~~~ - https://review.opendev.org/927972 (2023.1/antelope(ironic)) - https://review.opendev.org/927979 (2023.1/antelope(ironic-python-agent)) - https://review.opendev.org/927970 (2023.2/bobcat(ironic)) - https://review.opendev.org/927978 (2023.2/bobcat(ironic-python-agent)) - https://review.opendev.org/927968 (2024.1/caracal(ironic)) - https://review.opendev.org/927976 (2024.1/caracal(ironic-python-agent)) - https://review.opendev.org/927965 (2024.2/dalmatian(ironic)) - https://review.opendev.org/927974 (2024.2/dalmatian(ironic-python-agent)) - https://review.opendev.org/927969 (Bugfix/24.0 (ironic)) - https://review.opendev.org/927967 (Bugfix/25.0 (ironic)) - https://review.opendev.org/927966 (Bugfix/26.0 (ironic)) - https://review.opendev.org/927983 (Bugfix/9.12 (ironic-python-agent)) - https://review.opendev.org/927981 (Bugfix/9.13 (ironic-python-agent)) - https://review.opendev.org/927984 (Bugfix/9.9 (ironic-python-agent)) - https://review.opendev.org/927982 (Unmaintained/victoria(ironic)) - https://review.opendev.org/927980 (Unmaintained/wallaby(ironic)) - https://review.opendev.org/927977 (Unmaintained/xena(ironic)) - https://review.opendev.org/927975 (Unmaintained/yoga(ironic)) - https://review.opendev.org/927973 (Unmaintained/zed(ironic)) Credits ~~~~~~~ - Dan Smith from Red Hat (CVE-2024-44082) - Jay Faulkner from G-Research (CVE-2024-44082) - Julia Kreger from Red Hat (CVE-2024-44082) References ~~~~~~~~~~ - https://launchpad.net/bugs/2071740 - http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-44082 Notes ~~~~~ - For more information about the unadvisability of calling 'qemu-img info' on untrusted images, see CVE-2024-04467. - The ironic unmaintained/* branches will receive no point releases, but patches for them are provided as a courtesy. - The ironic-python-agent unmaintained/* branches will receive no point releases, nor are patches for them available. - The ironic and ironic-python-agent bugfix/* branches will receive no point releases, but patches for them are provided as a courtesy. -- Brian Rosmaita OpenStack Vulnerability Management Team https://security.openstack.org/vmt.html

1 0

[keystoneauth] OpenID Connect support for Onelogin broke Okta
by Pavlo Shchelokovskyy 04 Sep '24

04 Sep '24

Hi all, I'd like to bring your attention and discuss the following bug https://bugs.launchpad.net/keystoneauth/+bug/2078437 When onboarding Caracal release internally, we've noticed a regression in keystoneauth, namely that openstack CLI using v3OidcPassword auth no longer works with Okta provider. Okta returns the following error: HTTP 401 {"error":"invalid_request","error_description":"Cannot supply multiple client credentials. Use one of the following: credentials in the Authorization header, credentials in the post body, or a client_assertion in the post body." It looks like this is the aftermath of the following patch https://review.opendev.org/c/openstack/keystoneauth/+/881969 that presumably aimed to fix the Onelogin provider. I've prepared a patch that should fix both my Okta use case and, to my understanding, also address the original Onelogin concern, but I do not obviously have the use case and Onelogin access the original author of the patch had, and also it seems the author has changed jobs and is no longer active in OpenStack. Maybe somebody still has such an access to try and validate the original Onelogin use case using my patch? Or at least validate my conclusions I've outlined in the bug comments on why this should work and satisfy both cases: https://review.opendev.org/c/openstack/keystoneauth/+/927581 CC Graeme Moss who has ported that original Onelogin patch to Antelope and Zed and thus might have some additional perspective. Best regards, -- Dr. Pavlo Shchelokovskyy Principal Software Engineer Mirantis Inc www.mirantis.com

1 0