Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-11-28
17:04:30 stephenfin if you can get those done for tomorrow, I'll happily +2
17:06:01 sean-k-mooney i would have to look at how we generate the fake nodes but yes we proably coudl use the cpu_dedicated_set flags to restict it to just one numa node
17:06:20 sean-k-mooney im going to adress the other nits this evening. ill give that a try too
17:06:53 sean-k-mooney stephenfin: this is the test change right https://review.opendev.org/#/c/674072/8/nova/tests/functional/libvirt/test_pci_sriov_servers.py@533
17:09:24 stephenfin correct
17:09:38 stephenfin I don't think we consider the flags when generating the fake nodes
17:10:41 sean-k-mooney ill give it a try for the node generation we should not have too
17:11:06 sean-k-mooney but we should be looking at it for cpu pinning so provided i enable pinning in the flaovr or image it might work
17:12:08 sean-k-mooney i would need to look at how we do the cpu numbering too. to work out the correct mask/range but ya ill fix the other issues then ill give it a try
17:18:32 sean-k-mooney stephenfin: cool so jsut finished reviewing the comments they make sense so ill work on that this evening
19:16:11 openstackgerrit Merged openstack/nova stable/pike: Changing scheduler sync event from INFO to DEBUG https://review.opendev.org/669127
19:16:18 openstackgerrit Merged openstack/nova master: Mask the token used to allow access to consoles https://review.opendev.org/220622
23:21:36 openstackgerrit Merged openstack/nova master: block_device: Copy original volume_type when missing for snapshot based volumes https://review.opendev.org/694497
23:25:36 openstackgerrit Merged openstack/nova master: Sanity check instance mapping during scheduling https://review.opendev.org/683730
#openstack-nova - 2019-11-29
01:38:28 openstackgerrit Takashi NATSUME proposed openstack/nova master: Update keypairs in saving an instance object https://review.opendev.org/683043
06:21:02 openstackgerrit OpenStack Proposal Bot proposed openstack/nova master: Imported Translations from Zanata https://review.opendev.org/694717
08:36:05 openstackgerrit Brin Zhang proposed openstack/nova master: PoC: Support re-configure the delete_on_termination in server https://review.opendev.org/693828
09:57:46 openstackgerrit Balazs Gibizer proposed openstack/nova master: Make the binding:profile handling consistent in NeutronFixture https://review.opendev.org/696526
09:57:47 openstackgerrit Balazs Gibizer proposed openstack/nova master: Extend NeutronFixture to handle multiple bindings https://review.opendev.org/696246
10:00:46 openstackgerrit Balazs Gibizer proposed openstack/nova master: Do not mock setup net and migrate inst in NeutronFixture https://review.opendev.org/696247
10:02:16 openstackgerrit Balazs Gibizer proposed openstack/nova master: Move _get_request_group_mapping() to RequestSpec https://review.opendev.org/696541
10:03:47 openstackgerrit Balazs Gibizer proposed openstack/nova master: Move _update_pci_request_spec_with_allocated_interface_name https://review.opendev.org/696574
10:05:19 openstackgerrit Balazs Gibizer proposed openstack/nova master: Support live migration with qos ports https://review.opendev.org/695905
11:39:41 openstackgerrit Lee Yarwood proposed openstack/nova stable/train: block_device: Copy original volume_type when missing for snapshot based volumes https://review.opendev.org/696683
11:39:41 openstackgerrit Lee Yarwood proposed openstack/nova stable/train: DNM - Test booting from an encrypted volume snapshot https://review.opendev.org/696684
11:40:46 openstackgerrit Balazs Gibizer proposed openstack/nova stable/train: Mask the token used to allow access to consoles https://review.opendev.org/696685
11:44:55 openstackgerrit Lee Yarwood proposed openstack/nova stable/stein: block_device: Copy original volume_type when missing for snapshot based volumes https://review.opendev.org/696686
11:44:55 openstackgerrit Lee Yarwood proposed openstack/nova stable/stein: DNM - Test booting from an encrypted volume snapshot https://review.opendev.org/696687
12:28:51 openstackgerrit Lee Yarwood proposed openstack/nova master: WIP libvirt: Request and store the realpath of encrypted volumes from os-brick https://review.opendev.org/696693
13:04:47 openstackgerrit Boris Bobrov proposed openstack/nova master: Also enable iommu for virtio controllers and video in libvirt https://review.opendev.org/684825
13:04:47 openstackgerrit Boris Bobrov proposed openstack/nova master: Create a controller for qga when SEV is used https://review.opendev.org/693072
13:04:48 openstackgerrit Boris Bobrov proposed openstack/nova master: Switch to uses_virtio to enable iommu driver for AMD SEV https://review.opendev.org/696697
14:51:09 openstackgerrit Merged openstack/nova master: nova-net: Remove 'nova-network' binary https://review.opendev.org/686816
14:51:16 openstackgerrit Merged openstack/nova master: docs: Blast most references to nova-network https://review.opendev.org/686817
14:57:36 openstackgerrit Merged openstack/nova master: nova-net: Remove associate, disassociate network APIs https://review.opendev.org/686814
15:20:12 openstackgerrit Merged openstack/nova master: Add contributor doc for resize and cold migrate https://review.opendev.org/695609
15:22:35 openstackgerrit Merged openstack/python-novaclient master: doc: Update Testing document https://review.opendev.org/695995
15:42:39 amorin hey guys
15:44:19 openstackgerrit Martin Midolesov proposed openstack/nova master: Implementing graceful shutdown. https://review.opendev.org/666245
16:07:52 stephenfin sean-k-mooney: Thoughts? https://review.opendev.org/#/c/695416/2/nova/virt/libvirt/driver.py
16:14:08 sean-k-mooney why?
16:15:04 sean-k-mooney that could change how some devices are reported
16:15:28 sean-k-mooney we report a device as a PF if its capably of creating VF
16:15:34 sean-k-mooney not if they are allocated
16:15:43 sean-k-mooney so im thinking no
16:18:59 stephenfin Ahhh, yes, there would be backwards compatibility implications
16:19:01 stephenfin Didn't think of that
16:19:15 sean-k-mooney ya and i think its logically incorect
16:19:30 stephenfin oh?
16:19:33 sean-k-mooney the device type is based on capablity not configuration
16:20:04 stephenfin See the long commit message rewrite I just posted on that though
16:20:18 sean-k-mooney if the device report supprot for sriov via the pci config space feature mechanium then its a pf regardless of if you allcoate vfs
16:20:19 stephenfin I think having it configuration-driver is probably more sensible, tbh
16:20:55 stephenfin right, but the only reason we distinguish between them is to prevent people using the PF instead of a VF by accident, right?
16:21:20 stephenfin so you can whitelist all devices from a specific vendor and not have to worry about attaching a PF, losing all your VFs in the process
16:21:24 sean-k-mooney no
16:21:37 sean-k-mooney we did not used to make that disticntion
16:21:50 stephenfin speaking of, gibi, you around today? Any chance you could sanity check this and its predecessor https://review.opendev.org/#/c/694522/ (it's docs)
16:22:10 sean-k-mooney as i said the requiremetn to add device type was added when neutorn vnic_type=direct-pysical was added
16:22:24 sean-k-mooney you could use the pf via vnic_type=direct before
16:22:32 gibi stephenfin: added to my list
16:23:09 sean-k-mooney you could also use a device of type-pci with vnic_type=direct before
16:23:19 stephenfin gibi: Ta. Burnt through most of that live migration of BW-aware ports series today, btw. Just one or two todos on the last few patches
16:23:52 gibi stephenfin: saw the reviews, thank you for your time
16:23:57 stephenfin sean-k-mooney: Hmm, so np
16:24:00 stephenfin whoops
16:24:04 stephenfin gibi: np
16:24:27 stephenfin sean-k-mooney: Okay, I'd forgotten about that
16:24:28 stephenfin Darn
16:24:33 stephenfin We should document that in the code somewhere
16:24:48 sean-k-mooney it does not work that way anymore
16:24:57 openstackgerrit Vlad Gusev proposed openstack/nova stable/queens: Fix false ERROR message at compute restart https://review.opendev.org/696733
16:25:08 stephenfin wdym?
16:25:21 sean-k-mooney as i said the nova changes that added support for neutron vnic_type=direct-pyical
16:25:32 sean-k-mooney chanded how device_type works
16:25:45 sean-k-mooney and made it so vnic_type=direct wont select pf
16:25:57 sean-k-mooney and it might not be able to select type-PCI
16:26:38 sean-k-mooney before that change device_type was optional in the pci_request
16:27:00 stephenfin so we can't change this because it would mean you wouldn't be able to attach an interface using 'direct-physical' unless you'd enabled VFs on the host?
16:27:28 stephenfin to effectively work around that check
16:27:32 sean-k-mooney yes that would be one side effect
16:29:48 sean-k-mooney why what a really detailed and well writtne spec ... https://specs.openstack.org/openstack/nova-specs/specs/newton/implemented/sriov-pf-passthrough-neutron-port.html
16:32:31 stephenfin :D
16:32:56 stephenfin sean-k-mooney: lemme know if I said anything in https://review.opendev.org/#/c/695416/ that doesn't make sense
16:43:37 sean-k-mooney sure lets trade. does this make sense https://bugs.launchpad.net/nova/+bug/1853445/comments/5
16:43:37 openstack Launchpad bug 1853445 in OpenStack Compute (nova) "pci dev_type type-PCI passthrough failed" [Undecided,Invalid] - Assigned to Wei Hui (huiweics)
16:44:16 openstackgerrit Merged openstack/nova master: Force config drive in nova-next multinode job https://review.opendev.org/695431
16:44:22 openstackgerrit Merged openstack/nova master: Fix incorrect command examples https://review.opendev.org/695832
16:44:54 sean-k-mooney stephenfin: yes what you wrote makes sense
16:46:49 sean-k-mooney oh by the way i updated the functional test and have them working ill be pushing the update in the next 5 mins or so almost done
16:46:54 sean-k-mooney stephenfin: ^
16:47:22 sean-k-mooney the cpu_dedicate_set option worked perfectly to target the correct numa node in the tests
16:47:33 stephenfin perfect
16:47:41 stephenfin I'll review then
16:48:08 sean-k-mooney actully i did notice one thing that was odd but i need to look into that later it works for what we wanted however
16:48:11 stephenfin sean-k-mooney: what you wrote makes sense too
16:52:27 openstackgerrit sean mooney proposed openstack/nova master: support pci numa affinity policies in flavor and image https://review.opendev.org/674072
17:02:05 stephenfin sean-k-mooney: Got time to respin for the few final nits I pointed out? https://review.opendev.org/#/c/674072/
17:02:16 sean-k-mooney sure
17:02:24 openstackgerrit Merged openstack/nova master: compute: Use long_rpc_timeout in reserve_block_device_name https://review.opendev.org/693537

Earlier   Later