| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2026-07-27 | |||
| 16:20:19 | bauzas | but let's do this in the open discussion | |
| 16:20:25 | bauzas | #topic Open discussion | |
| 16:20:34 | bauzas | #item Support for Hostname-based Ceph Mon References | |
| 16:20:38 | bauzas | who wrote this ? | |
| 16:20:44 | Leo[m] | Hello, that was me | |
| 16:20:53 | Leo[m] | open to the idea of having a configuration that allows users to choose to use hostnames instead of IP addresses? Or, is there any other way that you would prefer to solve this problem? Could the mon addresses be read from nova.conf or ceph.conf instead? Thank you for your time! Leo | |
| 16:20:53 | Leo[m] | Hello, I hope everyone is doing well. My name is Leo Jenkins and I’m an engineer from Bloomberg. At the moment, ceph mon dump is used to get the IP addresses of the cloud’s Ceph mons. This makes it difficult to change the hardware or upgrade to IPv6. We have a loose commitment from the upstream Ceph community to add a —hostname flag to ceph mon dump to have it return the Ceph mon hostnames. If Ceph does this, would Nova be | |
| 16:21:36 | bauzas | Leo[m]: I don't think this should be a nova.conf option | |
| 16:22:36 | Leo[m] | Ok, sounds good. What do you recommend instead? | |
| 16:22:38 | sean-k-mooney | Leo[m]: so today there are 2 ways we interact with ceph | |
| 16:22:47 | sean-k-mooney | either vai the cidner connection details | |
| 16:22:52 | sean-k-mooney | or directly | |
| 16:22:58 | bauzas | that being said, we have a few ceph options https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.ceph_mount_point_base | |
| 16:23:01 | sean-k-mooney | and when we conenct directly we depend on the ceph.conf | |
| 16:23:10 | sean-k-mooney | ususlly under /etc/ceph | |
| 16:23:29 | sean-k-mooney | so if this was all handeeld internally via ceph.conf and the relevent cli/python bindings | |
| 16:23:42 | Leo[m] | https://github.com/openstack/nova/blob/9e1434d255f59f10c0fdf75c3fe6bebfa315df10/nova/storage/rbd_utils.py#L191 | |
| 16:23:42 | Leo[m] | This is the specific code in question by the way in case that is not clear | |
| 16:24:14 | sean-k-mooney | it woudl be transparent to nova when using image_type=rbd | |
| 16:24:42 | sean-k-mooney | so that the codepath that still uses the clint | |
| 16:25:26 | sean-k-mooney | but i beivle its finding the mons via the ceph.conf | |
| 16:25:42 | sean-k-mooney | i guess waht you really want is for us to have an option to use the mon fqdn | |
| 16:25:46 | sean-k-mooney | in teh rbd url | |
| 16:25:57 | sean-k-mooney | if that was supproted in qemu ectra | |
| 16:27:26 | sean-k-mooney | Leo[m]: i think this woudl likely need a spec or a specless blueprint | |
| 16:27:58 | sean-k-mooney | Leo[m]: but i dont think the ide of decopleing form the ip and allowing a hostnaem or fqdn woudl be problematic | |
| 16:28:21 | sean-k-mooney | we woudl just need to ensure we properly supprot that across the stack (libvirt,qemu,ceph client, nova) | |
| 16:28:41 | Leo[m] | Ok, good to know. I can create a blueprint and follow up in a future meeting | |
| 16:28:46 | tkajinam | I'm wondering if that option needs to be backend specific ? or can be common for all ceph clusters | |
| 16:29:27 | tkajinam | and probably we should also decide whether we should take care of migration between two nodes with old ceph / new ceph | |
| 16:29:32 | bauzas | let's discuss this once we have the blueprint but sure :) | |
| 16:29:35 | tkajinam | but that could be discussed further in spec/bp | |
| 16:29:39 | tkajinam | yeah | |
| 16:29:44 | bauzas | tkajinam: agreed with your concerns | |
| 16:29:45 | sean-k-mooney | tkajinam: that why i was tiniking about a spec | |
| 16:30:00 | sean-k-mooney | tkajinam: but we just approved supprot for updating the rbd urls on live mirgation | |
| 16:30:03 | sean-k-mooney | so that will help | |
| 16:30:17 | bauzas | let's ask Leo[m] to create a bp and then once he wants us to look at it as a specless BP, we can say whether we want a spec or not | |
| 16:30:38 | sean-k-mooney | but ya, currently the info for the client is coming form teh /etc/cepch/ceph.conf indirectly by the client finding it | |
| 16:30:44 | bauzas | Leo[m]: you know you're definitely too late for Hibiscuss timeframe and that will be an Indri feature ask ? | |
| 16:31:39 | Leo[m] | Yes, I think my team just wants to know whether this would be accepted by the community in the long term. There does not need to be an imminent release date | |
| 16:32:11 | bauzas | if so, that looks a good usecase, but we have some concerns about the design | |
| 16:32:20 | sean-k-mooney | long term i think yes we jsut need to figure out how to feature gate this and adress the upgrade impact ectra | |
| 16:32:42 | Leo[m] | Sounds good. Thank you for your help | |
| 16:32:54 | bauzas | nw | |
| 16:32:57 | bauzas | moving on | |
| 16:33:01 | bauzas | anything else ? | |
| 16:33:10 | gibi | - | |
| 16:33:21 | bauzas | if not, I'm more than happy to keep our meeting short :) | |
| 16:33:27 | bauzas | thanks all | |
| 16:33:30 | opendevmeet | Log: https://meetings.opendev.org/meetings/nova/2026/nova.2026-07-27-16.00.log.html | |
| 16:33:30 | opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/nova/2026/nova.2026-07-27-16.00.txt | |
| 16:33:30 | opendevmeet | Minutes: https://meetings.opendev.org/meetings/nova/2026/nova.2026-07-27-16.00.html | |
| 16:33:30 | opendevmeet | Meeting ended Mon Jul 27 16:33:30 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | |
| 16:33:30 | bauzas | #endmeeting | |
| 16:33:34 | gibi | thanks | |
| 16:33:35 | elodilles | thanks o/ | |
| 16:33:36 | lajoskatona | o/ | |
| 16:33:52 | fwiesel | Thanks o/ | |
| 16:33:59 | bauzas | that is it for myself running the nova meeting, Uggla is back next week :) | |
| 16:34:02 | bauzas | woohoo | |
| 16:34:09 | Leo[m] | Thank you /o | |
| 16:34:14 | tkajinam | thanks ! | |
| 16:35:31 | keerthi | https://review.opendev.org/c/openstack/nova/+/971299 | |
| 16:35:31 | keerthi | #info Looking for support on iothreads for blk-scsi , help in https://review.opendev.org/c/openstack/nova-specs/+/953940 | |
| 16:35:51 | opendevreview | Goutham Pacha Ravi proposed openstack/nova master: Support cold migration with virtiofs shares https://review.opendev.org/c/openstack/nova/+/989633 | |
| 16:46:02 | sean-k-mooney | keerthi that a backprot fo an unmerge dfeatrhe to the unmainted branch, we merged a diffent impeltion of it on master so that shoudl not be backported | |
| 17:06:30 | opendevreview | Merged openstack/nova master: policy: initialize enforcer before publishing it https://review.opendev.org/c/openstack/nova/+/997142 | |
| 17:43:13 | opendevreview | Merged openstack/placement master: Reproduce bug 2160721 https://review.opendev.org/c/openstack/placement/+/997362 | |
| 17:43:14 | opendevreview | Merged openstack/placement master: Check overall capacity before generating candidates https://review.opendev.org/c/openstack/placement/+/997617 | |
| 17:44:01 | gmaan | melwitt: I updated this as per your feedback, only pending comment is for revert_resize which is little tricky in this case, I replied in gerrit with details please check https://review.opendev.org/c/openstack/nova/+/996299 | |
| 19:22:41 | melwitt | gmaan: thanks will look | |
| 22:16:44 | gmaan | gibi: this is thread pool executor shutdown one, https://review.opendev.org/c/openstack/nova/+/998571 | |
| 23:07:29 | opendevreview | Goutham Pacha Ravi proposed openstack/nova master: Support cold migration with virtiofs shares https://review.opendev.org/c/openstack/nova/+/989633 | |
| #openstack-nova - 2026-07-28 | |||
| 07:48:53 | gibi | gmaan: ack | |
| 08:04:22 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2026.1: Check overall capacity before generating candidates https://review.opendev.org/c/openstack/placement/+/998874 | |
| 08:04:22 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2026.1: Reproduce bug 2160721 https://review.opendev.org/c/openstack/placement/+/998873 | |
| 08:05:04 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2025.2: Check overall capacity before generating candidates https://review.opendev.org/c/openstack/placement/+/998876 | |
| 08:05:04 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2025.2: Reproduce bug 2160721 https://review.opendev.org/c/openstack/placement/+/998875 | |
| 08:05:36 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2025.1: Check overall capacity before generating candidates https://review.opendev.org/c/openstack/placement/+/998878 | |
| 08:05:36 | opendevreview | Balazs Gibizer proposed openstack/placement stable/2025.1: Reproduce bug 2160721 https://review.opendev.org/c/openstack/placement/+/998877 | |
| 11:28:43 | kleini | Did anybody had a change to look on https://bugs.launchpad.net/nova/+bug/2161090 and its Gerrit review? | |
| 11:48:48 | kleini | sean-k-mooney, I think, it might not be necessary to leak OpenStack version or other details through SMBIOS. Just mentioning, VM is running on OpenStack should be enough, right? And then my patch allows to use same mechanisms for bootstrapping x86 and aarch64 VMs. | |
| 11:52:14 | sean-k-mooney | correct | |
| 11:52:23 | sean-k-mooney | kleini: we shoudl not be incluing the packing info | |
| 11:52:37 | sean-k-mooney | but enso that is not realy a correct fix | |
| 11:52:51 | sean-k-mooney | kleini: cloud-init shoudl not be using the smbios info to determin if tis openstack | |
| 11:53:02 | sean-k-mooney | that only worked for the libvirt driver | |
| 11:53:15 | sean-k-mooney | not for ironic or vmware or other nova virt drivers | |
| 11:53:39 | sean-k-mooney | kleini: also hte bug you filed is a duplcite of 2-3 others | |
| 11:54:22 | kleini | How does that work for Ironic, VMware or other Nova virt drivers? | |
| 11:54:24 | sean-k-mooney | https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/2069607 https://bugs.launchpad.net/cloud-init/+bug/1663304 | |
| 11:54:52 | sean-k-mooney | kleini: cloud-init is ment to use the metadata service well knwo ip to determin if its on openstack or aws | |
| 11:55:22 | sean-k-mooney | kleini: they fundemtally broke openstack suprpot with https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/2069607 | |
| 11:56:24 | sean-k-mooney | on an openstack cloud we have anti spoffing rules so that you cant impersonate http://169.254.169.254/openstack | |
| 12:10:47 | kleini | okay, so this needs to be fixed in cloud-init and until then I need to place a file in every image /etc/cloud.cfg.d to tell cloud-init, it is running on OpenStack. okay for me. | |
| 12:12:59 | gibi | fyi filed bug for the unstable tox-cover run https://bugs.launchpad.net/nova/+bug/2161963 I will push a fix soon | |
| 12:18:48 | opendevreview | Balazs Gibizer proposed openstack/nova master: [cover]stabilize test_migrate_from_disabled_host https://review.opendev.org/c/openstack/nova/+/998925 | |
| 12:18:54 | gibi | here you have it ^^ | |
| 13:46:01 | opendevreview | Balazs Gibizer proposed openstack/nova master: [cover]stabilize test_migrate_from_disabled_host https://review.opendev.org/c/openstack/nova/+/998925 | |