- openstack-discuss - lists.openstack.org

[swift][ptg] Ops Feedback Session - Th 10 Apr at 14:00 UTC
by Tim Burke 10 Apr '25

10 Apr '25

It's that time again! Let's get devs and ops together to talk about Swift: what's working, what isn't, and what would be most helpful to improve. We're meeting on meetpad (https://meetpad.opendev.org/p/swift-flamingo-ops-feedback) on Wednesday at 14:00UTC -- if you run a Swift cluster, we hope to see you there! Even if you can't make it, we'd appreciate it if you can offer some feedback on the etherpad (https://etherpad.opendev.org/p/swift-flamingo-ops-feedback) This is always=A0a highlight at every PTG for the swift developers. Have your say and help make the next Swift the best one yet! Tim

1 0

GPU PCI passthrough woes.
by Andy Speagle 10 Apr '25

10 Apr '25

Hey Folks, I could use a little assistance getting GPU passthrough working. I had this working already for one flavor of nvidia gpu... and I've added some hosts with a much newer gpu... I've updated the pci_alias and pci_passthrough variables and those seem to be getting set properly in nova.conf passthrough_whitelist = [{"vendor_id":"10de", "product_id":"1b06"},{"vendor_id":"10de", "product_id":"26b9"}] alias = {"name": "gpu", "product_id": "1b06", "vendor_id": "10de"} alias = {"name": "gpu-l40s", "product_id": "26b9", "vendor_id": "10de"} I believe I have all of the iommu stuff configured and have the pci- stub module entries... dmesg output shows that the GPUs are being claimed by the stub module. $ openstack flavor show t1.small_gpu_l40s +----------------------------+--------------------------------------+ | Field | Value | +----------------------------+--------------------------------------+ | OS-FLV-DISABLED:disabled | False | | OS-FLV-EXT-DATA:ephemeral | 0 | | access_project_ids | None | | description | None | | disk | 0 | | id | af70c94e-0026-4a39-bc1e-dfb93b286a54 | | name | t1.small_gpu_l40s | | os-flavor-access:is_public | True | | properties | pci_passthrough:alias='gpu-l40s:1' | | ram | 2048 | | rxtx_factor | 1.0 | | swap | 0 | | vcpus | 1 | +----------------------------+--------------------------------------+ Yet... I can't seem to get an instance to run using that new flavor... keeps complaining about there not being enough hosts available. Fault: code: 500 created: 2024-10-30T02:04:33Z message: "No valid host was found. There are not enough hosts available." details: | Traceback (most recent call last): File "/usr/lib/python3/dist-packages/nova/conductor/manager.py", line 1580, in schedule_and_build_instances host_lists = self._schedule_instances(context, request_specs[0], File "/usr/lib/python3/dist-packages/nova/conductor/manager.py", line 940, in _schedule_instances host_lists = self.query_client.select_destinations( File "/usr/lib/python3/dist- packages/nova/scheduler/client/query.py", line 41, in select_destinations return self.scheduler_rpcapi.select_destinations(context, spec_obj, File "/usr/lib/python3/dist-packages/nova/scheduler/rpcapi.py", line 160, in select_destinations return cctxt.call(ctxt, 'select_destinations', **msg_args) File "/usr/lib/python3/dist- packages/oslo_messaging/rpc/client.py", line 189, in call result = self.transport._send( File "/usr/lib/python3/dist- packages/oslo_messaging/transport.py", line 123, in _send return self._driver.send(target, ctxt, message, File "/usr/lib/python3/dist- packages/oslo_messaging/_drivers/amqpdriver.py", line 689, in send return self._send(target, ctxt, message, wait_for_reply, timeout, File "/usr/lib/python3/dist- packages/oslo_messaging/_drivers/amqpdriver.py", line 681, in _send raise result nova.exception_Remote.NoValidHost_Remote: No valid host was found. There are not enough hosts available. Traceback (most recent call last): File "/usr/lib/python3/dist- packages/oslo_messaging/rpc/server.py", line 241, in inner return func(*args, **kwargs) File "/usr/lib/python3/dist-packages/nova/scheduler/manager.py", line 223, in select_destinations selections = self._select_destinations( File "/usr/lib/python3/dist-packages/nova/scheduler/manager.py", line 250, in _select_destinations selections = self._schedule( File "/usr/lib/python3/dist-packages/nova/scheduler/manager.py", line 416, in _schedule self._ensure_sufficient_hosts( File "/usr/lib/python3/dist-packages/nova/scheduler/manager.py", line 455, in _ensure_sufficient_hosts raise exception.NoValidHost(reason=reason) nova.exception.NoValidHost: No valid host was found. There are not enough hosts available. Any clues on how/where to dig into this more to see what might be missing? Thanks. -- Andy Speagle

3 2

[nova][rfe] Support for memory allocation mode='immediate' in libvirt XML for large memory VMs
by 김동훈 09 Apr '25

09 Apr '25

1 0

[telemetry] Export Openstack Metrics to Prometheus
by Fernandes Figueiredo, Alex 09 Apr '25

09 Apr '25

Hello Openstack Telemetry Team, Do you know if we have a Prometheus exporter already available to export Openstack metrics stored by Gnocchi? I'm currently working on re-enabling the Openstack Telemetry services for the StarlingX Openstack (STX-O) project and we are moving in the direction of using the standard Openstack telemetry pipeline: openstack services -> ceilometer -> gnocchi -> aodh. However, to enable the integration with other systems and tools, we are looking for a way to export the Openstack metrics stored by Gnocchi to Prometheus. Thank you in advance for any help. Regards, Alex Figueiredo.

2 1

[all] OpenInfra Summit Europe Scheduled for October 2025
by Jeremy Stanley 08 Apr '25

08 Apr '25

In case anyone hasn't seen the news already, OpenInfra Summit Europe has been officially scheduled for October 17-19, 2025, at École Polytechnique near Paris, France: https://openinfra.dev/blog/openinfra-summit-2025 The OpenInfra Foundation is currently recruiting interested individuals to join a Programming Committee kickoff call on Thursday (2025-03-27). This call will be used to plan the conference tracks for the OpenInfra Summit ahead of the CFP, and will greatly benefit from OpenStack's involvement as always. If you would like to participate in the Programming Committee or assist in any other way, please email Helena Spease <helena(a)openinfra.dev>. I am also happy to try and point people in the right direction if you have any questions and prefer to reach out to me. On behalf of the foundation, we’re excited to bring the OpenInfra community together in France, and sincerely hope you'll join us! -- Jeremy Stanley

1 1

[cinder] 2025.2 PTG reminder and schedule
by Jon Bernard 08 Apr '25

08 Apr '25

Hello all, We covered the schedule in last week's meeting but I wanted to post it here for reference. For Cinder, we are meeting Wednesday, Thursday, and Friday of this week from 14:00 UTC until 16:00 UTC. Notable cross-projects sessions are: Glance/Cinder/Nova Wednesday 14h --> 15h UTC Neutron/Cinder/Nova Thursday 13h --> 14h UTC Please note that the Neutron/Nova session is the hour *before* our session block. As we did in the previous PTG, we'll be using the same google meet link and the sessions will be recorded. Everything should be noted in the etherpad. If you notice something is missing please add it or reach out with any questions. See you all in a couple of days! [1] https://etherpad.opendev.org/p/apr2025-ptg-cinder -- Jon

2 3

[neutron] Moving CI Meeting to Monday 1400 UTC
by Yatin Karel 08 Apr '25

08 Apr '25

Hello Neutron Team !! As discussed/agreed over the PTG, pushed the changes[1][2] for the new timing. Next Monday CI meeting will be at the new time. [1] https://review.opendev.org/c/opendev/irc-meetings/+/946631 [2] https://etherpad.opendev.org/p/neutron-ci-meetings -- Thanks and Regards Yatin Karel

1 0

security auditing and compliance in OpenStack environments
by acc.bu.mo＠gmail.com 08 Apr '25

08 Apr '25

Hi everyone, I'm currently working on my end-of-studies project focused on security auditing and compliance in OpenStack environments. As part of this project, I'm exploring how OpenStack deployments can align with compliance frameworks, especially SecNumCloud (from the French cybersecurity agency ANSSI). However, I'm also very interested in hearing about any tools, methods, or practices used for other standards (ISO 27001, CIS, FedRAMP, etc.). I'm still new to OpenStack and cloud compliance in general, and I'm finding it challenging to identify practical tools or workflows that companies actually use to perform security audits or compliance checks in OpenStack. If you're using (or have used) any tools, scripts, checklists, middleware, SIEM integrations, etc. for auditing OpenStack security or compliance I would be incredibly grateful if you could share your experience or point me in the right direction 🙏 Thank you so much in advance! Best regards,

2 1

Debian OpenStack Epoxy packages are available
by Thomas Goirand 08 Apr '25

08 Apr '25

Hi, Just a quick mail to tell everyone that I have finished Tempest validation of OpenStack Epoxy for OpenStack Bookworm backports. Currently, 1468 tests are passing, and there's only 8 tests that I need to investigate, though looking at the logs, it doesn't seem anything important is ongoing. Also note that I have integrated all Designate tests, that are all passing ok. As always, packages are available through osbpo.debian.net unofficial backports, accessible using extrepo. Under Bookworm, simply do: apt-get install extrepo extrepo enable openstack_epoxy apt-get update I haven't started validation under Debian 13 (aka: Trixie). I'm am currently working on it. It's going to be a little bit more difficult as Trixie runs RabbitMQ 4.0.5, which doesn't support HA queues anymore, so I need to switch to quorum queues in my test bed. Also, Trixie is running Python 3.13, and I've heard Eventlet is still not 100% compatible with it. Having Epoxy in Trixie is nice, because this means Bookworm users will be able to skip Dalmatian, and we're now aligned with the SLURP thing. As always, bug reports are welcome. Cheers, Thomas Goirand (zigo)

1 0

[Kolla-ansible]Dalmatian- How fix this bug: 'Port' object has no attribute 'security_groups
by Franck VEDEL (UGA) 08 Apr '25

08 Apr '25

Hi, New installation with kolla-ansible, Dalmatian, Ubuntu 2204. Everything works… except « Edit port » in Horizon. This bug is known: https://bugs.launchpad.net/horizon/+bug/2089557 My problem is: How to apply the fix with kolla-ansible. Can you help me? This bug is really a problem for my students who use Horizon. Thanks a lot Franck

1 0

[publiccloud-sig] Sessions during PTG
by Felix Kronlage-Dammers 08 Apr '25

08 Apr '25

Hi, the publiccloud-sig will meet twice as part of the PTG. On Tuesday (today) there will be a session at 16 UTC. On Wednesday (tomorrow) we will meet in our usual summertime slot at 7 UTC. The sessions will be here: https://meetpad.opendev.org/apr2025-ptg-public-cloud-sig (check also https://ptg.opendev.org/ptg.html) Looking forward to lively participation :) cheers felix -- Felix Kronlage-Dammers Open Source Business Alliance - Bundesverband für digitale Souveränität e.V. Tel.: +49-30-206539-205 | Matrix: @fkronlage:matrix.org | fkr(a)osb-alliance.com

1 0

[tacker] canceling weekly meeting
by Yasufumi Ogawa 08 Apr '25

08 Apr '25

Hi tacker team, Due to the vPTG, I'm canceling today's irc meeting. Yasufumi

1 0

[security-sig] Office hours during April 2025 PTG
by Jeremy Stanley 08 Apr '25

08 Apr '25

I've booked two PTG slots for OpenStack Security SIG office hours at times where hopefully at least one works no matter which timezone you live in. These days and times were also not heavily used by other teams, so should be fairly conflict-free: Tuesday (2025-04-08) at 13:00-14:00 UTC Thursday (2025-04-10) at 21:00-22:00 UTC There is no set agenda. I'll have the meetpad open for the full hour in case anyone wants to drop in with security-related topics on their mind. See you there! (Or hear you at least; I won't have a camera connected, you don't need one either.) -- Jeremy Stanley

1 0

[ironic] Call to action: Fix SNMP driver, or it may have to go
by Jay Faulkner 08 Apr '25

08 Apr '25

Hey all, So as many of you know, Ironic has had a driver for supporting SNMP-based PDUs (where you use an SNMP command to power on/off servers) for a while. This driver heavily uses pysnmp, a library created by a late Ironic contributor that was taken over by another company afterwards. This library is good, but is extremely fast moving and difficult for Ironic developers to keep up with. In fact, right now, both our SNMP-BMC emulator (virtualpdu) and Ironic are stuck using pysnmp-lexstudio, an old version of pysnmp which is now preventing other packages, e.g. pyasn1, from being upgraded. This is unsustainable in the long term, and it's been in this state for a while already. If you are an operator or organization that utilizes the SNMP driver, you need to rapidly find another method to manage power on your servers or help the Ironic developers maintain the SNMP driver -- at a minimum getting us onto the new pysnmp library (not a small undertaking; it's unexpectedly difficult). It's my intention to propose to the Ironic team that we mark this driver as unsupported and remove any remaining functional testing for it in anticipation of fully removing the driver in a future cycle. I do not want to take this action, but I'd rather have a driver removed from Ironic than one that's not going to work or hold the rest of the OpenStack requirements back. If you use this driver, please get in contact with the Ironic team if you have any resources to help with this migration. Feel free to use me-directly as a contact point if you're unsure where to start. Thanks, Jay Faulkner Ironic Security Liason

1 0

[neutron] PTG schedule created
by Brian Haley 07 Apr '25

07 Apr '25

Hello Neutrinos: I updated the Neutron Etherpad for the PTG to include day/time information for all proposed topics for the week [0]. Please make sure to check if I made any mistakes, or if you think we need more time for discussions. If necessary we can add time on Friday if we need it. Thanks, -Brian [0] https://etherpad.opendev.org/p/apr2025-ptg-neutron

1 0

[neutron] Neutron sriov vf-lag port binding
by Serhat Rıfat Demircan 07 Apr '25

07 Apr '25

Hello everyone, I have tested sriov with Connectx-5 cards, it works well but I also want to try ovs with hardware offload feature. I believe I have properly configured pf, vfs and bonding :). PF’s are working in switchdev mode (I have followed stackhpcs article[1]). Unfortunately, I see port binding errors when creating instances. I have created a vlan provider network and a port with vnic_type=direct under that network. I am trying to create instance with that port. nova-compute.log > 2025-04-04 17:37:28.014 7 ERROR nova.compute.manager [instance: > bf8408e8-20a9-4963-9e50-033e54aba9fe] nova.exception.PortBindingFailed: > Binding failed for port 6cbd8181-5707-4349-9fa0-c35e7370bfcd, please check > neutron logs for more information. neutron-server.log > 025-04-04 17:37:27.000 25 DEBUG > neutron.plugins.ml2.drivers.mech_sriov.mech_driver.mech_driver > [req-ff074be0-e410-4df8-8938-7e2b59a84bfd > req-d9bffb08-7e74-4976-b3ae-8cb544ec71f5 fbbdf9e30cad498180c3b637d1bba6d9 > 3e7a1867f1c24a158b4bb2d2f3f68b9c - - default default] Refusing to bind due > to unsupported vnic_type: direct with switchdev capability bind_port > /var/lib/kolla/venv/lib/python3.10/site-packages/neutron/plugins/ml2/drivers/mech_sriov/mech_driver/mech_driver.py:130 neutron/plugins/ml2/drivers/mech_sriov/mech_driver/mech_driver.py > if (vnic_type == portbindings.VNIC_DIRECT and > 'switchdev' in capabilities): > LOG.debug("Refusing to bind due to unsupported vnic_type: %s " > "with switchdev capability", > portbindings.VNIC_DIRECT) nova.conf > [pci] > passthrough_whitelist = [{"physical_network": "physnet5", "devname": > "ens10f0"},{"physical_network": "physnet6", "devname": "ens10f1"}] [sriov_nic] physical_device_mappings = physnet5:ens10f0,physnet6:ens10f1 Is vnic_type=direct wrong if switchdev in capabilities or is this a configuration issue? Environment: Openstack version: Caracal Operating System: Ubuntu 22.04 Thanks, Serhat [1] - https://www.stackhpc.com/vflag-kayobe.html

3 3

[neutron] Bug deputy week 14 (March 31 - April 6)
by Rodolfo Alonso Hernandez 07 Apr '25

07 Apr '25

Hello Neutrinos: This is the bug list of week 14 (March 31 - April 6). HIGH: * https://bugs.launchpad.net/neutron/+bug/2105977 Random error in ``test_packets_of_any_connection_state_can_reach_dest`` test. Not assigned yet. * https://bugs.launchpad.net/neutron/+bug/2105502 service role permissions not enough for octavia allowed address pair driver. Assigned. Two patches proposed: https://review.opendev.org/c/openstack/neutron/+/861169 and https://review.opendev.org/c/openstack/neutron/+/945329. MEDIUM: * https://bugs.launchpad.net/neutron/+bug/2106231 [OVN][QoS] The Logical_Swith_Port QoS options are removed when the port is updated. Assigned. Patch proposed: https://review.opendev.org/c/openstack/neutron/+/946542 RFE: * https://bugs.launchpad.net/neutron/+bug/2105855 [RFE] VXLAN physical_networks or new network type Regards.

1 0

[manila]NFS-Ganesha exports
by Mikhail Okhrimenko 06 Apr '25

06 Apr '25

Hello community, Trying to setup the Manila version of Caracal with NFS-Ganesha backend. Ganesha version 4.0 is deployed using ceph orch in a ceph cluster version 17. Backend configuration in manila.conf [ganesha_nfs] driver_handles_share_servers = False share_backend_name = ganesha_nfs Share_driver = manila.share.drivers.cephfs.driver.CephFSDriver cephfs_protocol_helper_type = NFS cephfs_conf_path = /etc/ceph/ceph.conf cephfs_auth_id = manila cephfs_cluster_name = ceph cephfs_filesystem_name = {{ cephfs_filesystem_name }} cephfs_nfs_cluster_id = {{ cephfs_nfs_cluster_id }} I see the subvolume being created in ceph ceph fs subvolume ls <volume_name> but no export ceph nfs export ls <cluster_id> what am I doing wrong? what am I missing? in the manila-share logs I see a request to create a subvolume, but I don't see a request to create an export, for some reason manila doesn't even try to create an export

3 7

bug report
by Xavier Benjamin 05 Apr '25

05 Apr '25

whenever i try to connect to trove pts servers it would usually say fails to authenticate or something. Can you fix that please?

2 2

[all][tc][ptg] OpenStack TC Flamingo PTG Schedule, Etherpad and Weekly Meeting updates
by Goutham Pacha Ravi 05 Apr '25

05 Apr '25

Hello Stackers, The OpenStack Technical Committee has scheduled discussions at the following time slots next week: - April 7 2025: 1300 UTC to 1500 UTC - April 11 2025: 1500 UTC to 1700 UTC You can find the discussion times and meeting details on this etherpad: https://etherpad.opendev.org/p/apr2025-ptg-os-tc Of special importance is the TC/Community Leaders forum session on 2025-04-11 at 1500 UTC. The list of topics for this forum session can expand over the week. If you're a maintainer on any OpenStack repository, or an aspiring maintainer, please do join us there. The etherpad, the schedule and a link to the meeting room can also be found on the PTG website [1]. At this time, if you'd like to add more topics, please reach out to me, or chime-in on the #openstack-tc channel on OFTC. In lieu of the PTG, we'll not have the regular weekly IRC meeting next week. The IRC meeting will be back on 2025-04-15 [2]. Thank you, Goutham Pacha Ravi (gouthamr) OpenStack TC Chair [1] https://ptg.opendev.org/ptg.html [2] https://wiki.openstack.org/wiki/Meetings/TechnicalCommittee

1 0

[nova] vGPU not attached to VM using Cyborg with OpenStack Caracal
by ts.jung＠okestro.com 04 Apr '25

04 Apr '25

Hello, I'm facing an issue with vGPU assignment through Cyborg in OpenStack Caracal. Although Cyborg reports that the ARQ is bound successfully, the vGPU device is not attached to the VM in Nova. - Environment: OS: Ubuntu 22.04.3 LTS Kernel: 5.15.0-78-generic Nova API: 29.2.0 Cyborg API: 14.1.0.dev1 GPU: NVIDIA A100 PCIe 40GB - Device Profile: openstack accelerator device profile list +--------------------------------------+------+-----------------------------------------------------------------------------+ | uuid | name | groups | +--------------------------------------+------+-----------------------------------------------------------------------------+ | 8425925f-7a3c-4336-8a1c-270f481a28ae | vgpu | [{'resources:VGPU': '1', 'trait:CUSTOM_NVIDIA_20F1_A100_1_5C': 'required'}] | +--------------------------------------+------+-----------------------------------------------------------------------------+ Flavor: openstack flavor show vgpu --column properties +------------+-----------------------------+ | Field | Value | +------------+-----------------------------+ | properties | accel:device_profile='vgpu' | +------------+-----------------------------+ Result After VM Creation: openstack accelerator arq list +--------------------------------------+------+-----------------------------------------------------------------------------+ | uuid | name | groups | +--------------------------------------+------+-----------------------------------------------------------------------------+ | 8425925f-7a3c-4336-8a1c-270f481a28ae | vgpu | [{'resources:VGPU': '1', 'trait:CUSTOM_NVIDIA_20F1_A100_1_5C': 'required'}] | +--------------------------------------+------+-----------------------------------------------------------------------------+ ARQ shows Bound with an attach_handle_type of MDEV, but the GPU device is not visible inside the VM. mdevctl list # Shows UUID and mdev type virsh dumpxml instance-xxxx # No mdev device listed Logs (Summary) Cyborg logs indicate ARQ binding was successful and attach handle allocated (MDEV). Nova-compute logs show: Ignoring accelerator requests for instance ... Supported Attach handle types: {'PCI'}. But got these unsupported types: {'MDEV'}. Question. Is there something I’m missing in the integration between Cyborg and Nova for MDEV attach_handle_type? It seems like Nova is not yet supporting MDEV, even though Cyborg bound the ARQ correctly. Any suggestions or known workarounds would be appreciated. Thank you!

2 1

[i18n] i18n SIG - 2025.2 (Flamingo) Virtual PTG Schedule
by Ian Y. Choi 04 Apr '25

04 Apr '25

Hi, I would like to share the i18n SIG schedule for the upcoming PTG. Dates and locations are: • April 9th, 1400 ~ 1500 UTC at kilo-WedB2 (Please find https://ptg.opendev.org/ptg.html for real time information) Note that I18n PTG Etherpad link is: https://etherpad.opendev.org/p/apr2025-ptg-i18n . Looking forward to the meeting and interactions during the PTG! Thank you, /Ian

1 0

[Octavia] Active/Standby with AZ
by Vincent Godin 04 Apr '25

04 Apr '25

Hello, To address the Octavia and Cinder backend mismatch, I'm going to temporarily create a Cinder backend stretched across all three AZs. Now I have another question: is it possible to use Octavia AZs to set up an active/standby load balancer so that the active load balancer is on one AZ and the standby load balancer is on another one ? Of course, the network is stretched across all three AZs, so the VRRP mechanism is possible. Vincent

3 2

Neutron
by Bowley, Jonathan T 04 Apr '25

04 Apr '25

Hey I've been struggling with trying to get my kolla-ansible openstack multinode deployment working with my external trunked port i have openstack connected to and also using my external dhcp server. Does anyone have any thoughts on what I could be missing? I grasping at straws at this point and ill buy ya dinner if you can help me! when I launch a vm i see it assigning vms an ip but its no way its coming from my external dhcp server i think its just coming from its own pools. Also to add I'm using 2024.2 My global yaml - enable_neutron_provider_networks: "yes" neutron_external_interface: "bond0" network_interface: "eno3" when running ip a i see which i have no clue if they are suppose to say down in my head it doesn't seem right but im not sure because i havent had a successful deployment yet so not sure what its suppose to look like. bond0 <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue master ovs-system state UP group default qlen 1000 ovs-system ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000 br-ex: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000 br-int: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000 br-tun: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000 extra conf files /etc/kolla/config/neutron/ml2_conf.ini [ml2] type_drivers = vlan tenant_network_types = vlan mechanism_drivers = openvswitch extension_drivers = port_security [ml2_type_vlan] network_vlan_ranges = physnet1:100:100,physnet1:144:144,physnet1:513:513 /etc/kolla/config/neutron/openvswitch_agent.ini [ovs] bridge_mappings = physnet1:br-ex Thanks, Jonathan Bowley IT Specialist / DevOps Engineer / SysAdmin IT Services National Geospatial Technical Operations Center (NGTOC<http://ngtoc.usgs.gov/>) U.S. Geological Survey Rolla, Missouri

2 1

Non predictable device name/path in libvirt/qemu config with multipath
by Gilles Mocellin 03 Apr '25

03 Apr '25

Hello Stackers, Today I had a stressful problem. While doing system upgrades on my Ubuntu based OpenStask-Ansible deployment, with Pure Storage cinder driver and boot from volume instances, I faced that problem : While migrating instances to evacuate a compute node, several instances never wanted to live migrate (Kubernetes workers with 16GB RAM). After restarting several times, attempting to change some nova/libvirt configuration (certainly not taken, because qemu was not restarted on source host), I decider to cold migrate thoses instances. So I simply use : openstack server migrate instance-uuid The instances tried to start on other compute node, but failed. Two have the same problem I would talk about, the third just didn't find the volume/LUN on the storage Array (also very weird). So, why the other two didn't start ? They wanted to use a multipath map that was already in use. And really in use from another instance. the multipath -f mpathxy was tried and make the multipath map unusable for the instance which used it. It was the database of an important app, to add a little bit of stress. This is not happening with successful live migration, I only see it with cold migrations. But with some instances, I really don't manage to have a successful live migration, so it can be useful to also have a working cold migration (I mean stop, migrate, and start the instance). So, my questions are : 1. Should multipath.conf enable or not user friendly names ? I didn't find a recommended multipath config. I saw both in kolla templates (not using friendly names) or starlingX ones or old bugs (using friendly names, for data flush to really be effective on paths...) Didn't find yet if openstack-ansible is managing multipath or not. I would say that without friendly names, we have predictable names in /dev/ mapper/ (with WWN), that if used couldn't lead to device "steal" if every instance record that predictable device. 2. Is there some configuration for libvirt/qemu ? Because, in the libvirt/qemu instance config file, the devices are not the multipath map but the /dev/dm-XY where the multipath map symlinks to... Also not predictable and changing, especially when changing compute hosts with different number of accessible volumes... Did someone already had those problems, and how do you handled it ?

1 0