| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-08-29 | |||
| 07:39:28 | opendevreview | Balazs Gibizer proposed openstack/nova master: Test multi create with PCI in placement https://review.opendev.org/c/openstack/nova/+/854663 | |
| 07:39:28 | opendevreview | Balazs Gibizer proposed openstack/nova master: Test reschedule with PCI in placement https://review.opendev.org/c/openstack/nova/+/854626 | |
| 07:39:30 | opendevreview | Balazs Gibizer proposed openstack/nova master: DNM: Enable pci in placement feature https://review.opendev.org/c/openstack/nova/+/854925 | |
| 07:39:30 | opendevreview | Balazs Gibizer proposed openstack/nova master: Allow enabling PCI scheduling in Placement https://review.opendev.org/c/openstack/nova/+/854924 | |
| 08:29:05 | songwenping | gibi: hi, can we delete the server group policy that is used by the vms? | |
| 08:31:07 | sean-k-mooney | songwenping: not really, you may be able to delete the server group but you will not be able to add those vms to anohter | |
| 08:31:46 | sean-k-mooney | https://docs.openstack.org/api-ref/compute/?expanded=delete-server-group-detail#delete-server-group | |
| 08:33:02 | songwenping | sean-k-mooney: thanks, but this will effect the vm's evacuate and migrate if delete. | |
| 08:33:09 | sean-k-mooney | songwenping: there is an api to delete the server group but just be aware that your exiting instances cand be added to a new server group in the future | |
| 08:33:47 | sean-k-mooney | if you delete the server group all affinity polices associated will be lost | |
| 08:35:23 | songwenping | why donnot we limit the delete? | |
| 08:35:41 | sean-k-mooney | limit the delete? | |
| 08:35:56 | sean-k-mooney | server groups are a normal tenant concept | |
| 08:36:08 | sean-k-mooney | so a project member should be able to create and or delete them | |
| 08:36:26 | sean-k-mooney | why would we limit the delete | |
| 08:36:58 | songwenping | no, if there are vms associated, we cannot delete the group. | |
| 08:37:32 | sean-k-mooney | thats really a teanat choice | |
| 08:37:37 | sean-k-mooney | we could block it | |
| 08:38:33 | sean-k-mooney | but we generally assume our tenant and oeprator are compitent by default | |
| 08:38:48 | sean-k-mooney | which is why we do not block delete falvors that are in user or images | |
| 08:39:54 | sean-k-mooney | deleteing an image potentially make a vm un evacuatatble | |
| 09:11:44 | songwenping | sean-k-mooney: thanks, got it. i have another question: on a customer's AMD server, the gpu's iommu group have many other devices which effects the passthrough to vm, how can we indivial the imommu group? | |
| 09:18:05 | opendevreview | Ghanshyam proposed openstack/nova master: Add documentation and releasenotes for RBAC change https://review.opendev.org/c/openstack/nova/+/854882 | |
| 09:25:22 | sean-k-mooney | songwenping: did they set amd_iommu=on | |
| 09:25:39 | songwenping | yes | |
| 09:26:05 | sean-k-mooney | then they have 3 possiable path forward | |
| 09:26:36 | sean-k-mooney | 1 move the card to a pci slot that is directly connected to the cpu and not the chipset | |
| 09:26:39 | songwenping | when i set all the devices' driver to vfio-pci, the gpu can passthrough to the vm. | |
| 09:27:13 | sean-k-mooney | 2 check if the bios has the ablity to seperate the devie explictly | |
| 09:27:35 | sean-k-mooney | 3 use the out of tree kernel patch for iommu remapping | |
| 09:28:27 | sean-k-mooney | "PCIe ACS Override for bypassing IOMMU groups support." https://github.com/xanmod/linux-patches/blob/master/linux-5.19.y-xanmod/pci_acso/0001-pci-Enable-overrides-for-missing-ACS-capabilities.patch | |
| 09:30:11 | sean-k-mooney | songwenping: yes that a limitation of vfio-pci | |
| 09:30:21 | sean-k-mooney | "PCIe ACS Override for bypassing IOMMU groups support." https://github.com/xanmod/linux-patches/blob/master/linux-5.19.y-xanmod/pci_acso/0001-pci-Enable-overrides-for-missing-ACS-capabilities.patch | |
| 09:31:34 | sean-k-mooney | would allow you to force the device into its own iommu group if they cant take the simpler options of movign the card to be driectly conneccted to the cpu instead of the chipset | |
| 09:31:43 | songwenping | this patch is relatively new. | |
| 09:31:59 | sean-k-mooney | that version of it | |
| 09:32:09 | sean-k-mooney | the patch has been around for a few years | |
| 09:32:17 | sean-k-mooney | the orginal one is form 2018 | |
| 09:32:36 | sean-k-mooney | correct 2013 https://lkml.org/lkml/2013/5/30/513 | |
| 09:33:09 | sean-k-mooney | i honeestly dont know why this has not landed other then alex is busy with other things | |
| 09:33:17 | sean-k-mooney | it would be really really useful for vms | |
| 09:35:29 | songwenping | thanks, i will try to merge the patch. | |
| 09:36:03 | gibi | sean-k-mooney: I've reviewed the user_data and the rebuild bfv patches. I will go and dig into the func test part ofthe bfv patch now. | |
| 09:36:56 | gibi | sean-k-mooney, stephenfin: if you have cycles then the next 11 PCI patch is ready to land ;) | |
| 09:39:09 | opendevreview | Amit Uniyal proposed openstack/nova stable/victoria: add regression test case for bug 1978983 https://review.opendev.org/c/openstack/nova/+/854979 | |
| 09:39:10 | opendevreview | Amit Uniyal proposed openstack/nova stable/victoria: For evacuation, ignore if task_state is not None https://review.opendev.org/c/openstack/nova/+/854980 | |
| 09:57:39 | admin1 | hi .. i am getting this error: nova-compute Connection to libvirt failed: authentication failed: authentication failed: libvirt.libvirtError: authentication failed: authentication failed and the compute node is unable to register itself .. i am using kolla-ansible yoga .. | |
| 10:14:08 | sean-k-mooney | they changed how they connect | |
| 10:14:29 | sean-k-mooney | that is really a question for the kolla channel since nova was not involved in this change | |
| 10:14:50 | sean-k-mooney | admin1: they enabled more authentication by default | |
| 10:15:42 | sean-k-mooney | admin1: you can tempoarly disable ths use of sasl authentication https://github.com/openstack/kolla-ansible/blob/master/doc/source/reference/compute/libvirt-guide.rst#sasl-authentication | |
| 10:17:01 | sean-k-mooney | noonedeadpunk: was i chating to you about ^ in the past | |
| 10:17:09 | sean-k-mooney | triging to rememebr who hit that before | |
| 10:18:47 | sean-k-mooney | noonedeadpunk: porably wrong user name then | |
| 10:19:13 | sean-k-mooney | im trying to think who works at blizzard | |
| 10:19:37 | opendevreview | Amit Uniyal proposed openstack/nova stable/victoria: add regression test case for bug 1978983 https://review.opendev.org/c/openstack/nova/+/854979 | |
| 10:19:40 | sean-k-mooney | its been a month or two | |
| 10:19:40 | noonedeadpunk | we never did sasl in osa, we use tls certs auth between computes for migrations | |
| 10:19:43 | admin1 | maybe spatel :) | |
| 10:19:59 | sean-k-mooney | noonedeadpunk: tls is more secure for sure | |
| 10:20:00 | noonedeadpunk | afaik he's not from blizzard ) | |
| 10:20:13 | sean-k-mooney | spatel is not no | |
| 10:20:14 | admin1 | yeah | |
| 10:20:39 | sean-k-mooney | i could proably gerp my logs but admin1 that is likely your issue | |
| 10:20:42 | noonedeadpunk | btw talking about auth, it's sad that for offline migrations you still have to ensure SSH connection between computes :( | |
| 10:20:59 | sean-k-mooney | offline you mean cold | |
| 10:21:06 | noonedeadpunk | yeah, sorry, cold | |
| 10:21:26 | sean-k-mooney | ya for the disk copy and more importantly if you are using cpeh/bfv for the file system check | |
| 10:21:30 | noonedeadpunk | as for live migrations libvirt is jsut leveraged | |
| 10:21:34 | sean-k-mooney | that could be remvoed | |
| 10:22:47 | noonedeadpunk | can it? hm, I can recall I still had to ensure ssh even for ceph | |
| 10:22:53 | sean-k-mooney | but it need some changes to how we check for shared filesystem | |
| 10:23:38 | sean-k-mooney | we could replcase the ssh requirement with doign the check via an rpc | |
| 10:23:54 | sean-k-mooney | right now we ssh to the dest and toch a file | |
| 10:23:58 | noonedeadpunk | ah, ok, now I got what you meant | |
| 10:24:00 | sean-k-mooney | then check on the source if we can see it | |
| 10:24:08 | sean-k-mooney | but we could do that via an rpc | |
| 10:24:29 | noonedeadpunk | yeah, I thought the same, but ENOTIME as usual | |
| 10:25:42 | sean-k-mooney | for live migration we actully have the rpcs in place we woudl just do it avia the existing pre-migration rpcs | |
| 10:26:07 | sean-k-mooney | i dont know if we can do the same for cold migratoin today as i dont know if we have rpcs to the dest at the right times that we can extend | |
| 10:38:12 | opendevreview | ribaudr proposed openstack/nova master: Attach Manila shares via virtiofs (objects) https://review.opendev.org/c/openstack/nova/+/839401 | |
| 10:38:13 | opendevreview | ribaudr proposed openstack/nova master: Attach Manila shares via virtiofs (drivers and compute manager part) https://review.opendev.org/c/openstack/nova/+/833090 | |
| 10:38:13 | opendevreview | ribaudr proposed openstack/nova master: Attach Manila shares via virtiofs (manila abstraction) https://review.opendev.org/c/openstack/nova/+/831194 | |
| 10:38:14 | opendevreview | ribaudr proposed openstack/nova master: Bump compute version and check shares support https://review.opendev.org/c/openstack/nova/+/850499 | |
| 10:38:14 | opendevreview | ribaudr proposed openstack/nova master: Attach Manila shares via virtiofs (api) https://review.opendev.org/c/openstack/nova/+/836830 | |
| 10:38:15 | opendevreview | ribaudr proposed openstack/nova master: Add instance.share_attach notification https://review.opendev.org/c/openstack/nova/+/850501 | |
| 10:38:15 | opendevreview | ribaudr proposed openstack/nova master: Add metadata for shares https://review.opendev.org/c/openstack/nova/+/850500 | |
| 10:38:16 | opendevreview | ribaudr proposed openstack/nova master: Add instance.share_detach notification https://review.opendev.org/c/openstack/nova/+/851028 | |
| 10:38:17 | opendevreview | ribaudr proposed openstack/nova master: Add shares to InstancePayload https://review.opendev.org/c/openstack/nova/+/851029 | |
| 10:38:18 | opendevreview | ribaudr proposed openstack/nova master: Add instance.power_off_error notification https://review.opendev.org/c/openstack/nova/+/852278 | |
| 10:38:18 | opendevreview | ribaudr proposed openstack/nova master: Add instance.power_on_error notification https://review.opendev.org/c/openstack/nova/+/852084 | |
| 10:38:20 | opendevreview | ribaudr proposed openstack/nova master: Add libvirt test to ensure metadata are working. https://review.opendev.org/c/openstack/nova/+/852086 | |
| 10:38:20 | opendevreview | ribaudr proposed openstack/nova master: Add helper methods to attach/detach shares https://review.opendev.org/c/openstack/nova/+/852085 | |
| 10:38:22 | opendevreview | ribaudr proposed openstack/nova master: Add share_info parameter to reboot method for each driver (driver part) https://review.opendev.org/c/openstack/nova/+/854823 | |
| 10:38:22 | opendevreview | ribaudr proposed openstack/nova master: Add virt/libvirt error test cases https://review.opendev.org/c/openstack/nova/+/852087 | |
| 10:38:24 | opendevreview | ribaudr proposed openstack/nova master: Change microversion to 2.93 https://review.opendev.org/c/openstack/nova/+/852088 | |
| 10:38:24 | opendevreview | ribaudr proposed openstack/nova master: Support rebooting an instance with shares (compute and API part) https://review.opendev.org/c/openstack/nova/+/854824 | |
| 10:53:27 | opendevreview | Amit Uniyal proposed openstack/nova stable/victoria: add regression test case for bug 1978983 https://review.opendev.org/c/openstack/nova/+/854979 | |
| 10:53:28 | opendevreview | Amit Uniyal proposed openstack/nova stable/victoria: For evacuation, ignore if task_state is not None https://review.opendev.org/c/openstack/nova/+/854980 | |
| 10:54:53 | sean-k-mooney | gibi: did you fiture out why you could not do request_id=uuidutils.generate_uuid(dashed=True) | |