Friday, 2019-03-01

dansmithmriedem: melwitt Kevin_Zheng: sent a reply to the list00:03
*** wolverineav has quit IRC00:03
*** macza has quit IRC00:04
*** wolverineav has joined #openstack-nova00:04
Kevin_Zhengdansmith: cool, thanks, will check it later00:05
dansmithKevin_Zheng: no, you will check it now!00:06
dansmithI mean, of course :P00:06
mriedemKevin_Zheng: i left some comments in that functional test00:07
mriedemKevin_Zheng: i'd like to see a functional test that exercises that when attaching a new root volume with a new image, we properly update the request spec to ensure we unshelve on a host that only works for that new image00:08
*** wolverineav has quit IRC00:08
melwittI definitely sympathize with the idea of not wiping the tag on detach. that was my worry, was someone not realizing that would happen00:09
melwittI need more brain to think about the rest of it right now00:10
*** slaweq has joined #openstack-nova00:11
*** munimeha1 has quit IRC00:11
mriedemwhere the virus lives00:12
mriedemnice00:12
*** itlinux_ has quit IRC00:13
*** mvkr has joined #openstack-nova00:14
*** slaweq has quit IRC00:15
Kevin_Zhengmriedem: I will try to write one today00:17
*** itlinux has joined #openstack-nova00:17
melwitta virus? geez kevin00:17
*** tosky has quit IRC00:18
artommriedem, I don't even know anymore. I saw we revert device tagging altogether, who needs it anyways?00:19
artom*say00:19
Kevin_Zhengmelwitt: I will try my best:)00:21
Kevin_ZhengCheck the news tomorrow00:21
melwitt:D00:22
mriedemtrump told us we can't trust those huawei guys00:23
artomOut of curiosity, do we have a way to scope CONF objects (fixtures?) to a single service in func tests - specifically, I want 2 computes with different vcpu_pin_set (for example)00:24
mriedemdansmith: Kevin_Zheng: replied with some questions for clarification - i know it's getting late so probably have to pick this up tomorrow00:24
mriedemKevin_Zheng: regardless of the tag decision, my comments on the functional tests are probably enough to keep you busy for a bit00:24
mriedemartom: since CONF is global changes are no00:25
mriedem*chances00:25
melwittartom: I _think_ you can do that by self.flags(value for compute 1) start compute service 1, self.flags(value for compute 2), start compute service 200:25
mriedemthere is a ConfFixture, and fixtures are context managers,00:25
mriedemso it's possible you could temporarily set config via context00:26
melwittbut maybe it won't stick, if it's going to look at it after startup00:26
mriedemi'm going to be in big trouble for not getting dinner started so i have to run00:26
*** mriedem is now known as mriedem_away00:26
melwittyeah, maybe what I said would only work if it's a conf value only considered during init_host00:26
artomReason I'm asking is - without this, I don't see a way to func test NUMA LM meaningfully00:27
artommelwitt, that'd be enough, I tihnk00:27
dansmithmelwitt: only if the values are read and cached at start00:27
dansmithwhen I wrote the  oschown thing, I chased that down00:27
dansmithit's a huge problem00:27
melwittyeah, I think it won't work if the conf option is read later on00:27
dansmithin that it's completely fundamentally global as heck forever00:27
melwittbut I know I've done something like this before in func test where it was startup only00:27
melwitthaha yeah00:27
dansmithmelwitt: right, if it's read at startup and ever again, and it does that before the start returns, then it'll work, but it's fragile00:28
melwittyeah. definitely not saying it's robust00:28
artomBleah, vpcu_pin_set is read after startup00:28
melwittit's the kind of janky thing I do in tests when I find them00:28
melwittafter the desperation sets in00:29
artomdansmith, so, we can follow up on ML, but I think that rules out NUMA LM func testing?00:30
artomI don't see another way to "force a conflict", so to speak00:30
*** dpawlik has joined #openstack-nova00:30
dansmithartom: it means you have to mock something out, which we try to avoid in functional tests, but can't really avoid for hardware things (like virt driver stuff)00:31
dansmithwhich is why functional is not enough00:31
*** wolverineav has joined #openstack-nova00:31
artomdansmith, mock to different values with 2 compute services running?00:31
dansmithartom: mock somewhere other than config00:31
dansmithartom: we have useful functional tests for other things that depend on global state00:32
dansmithit's not trivial to do, but it's doable00:32
artomdansmith, can I have an example? I'll use that as inspiration :)00:32
dansmithartom: the cells stuff all depends on a completely global DB state00:33
dansmithit's been a challenge to get that stuff working because it's so global, but like numa, it's hard to test multiple cells for real, so they're quite useful00:33
artomdansmith, NUMA is a matter of hardware00:34
*** zhubx has quit IRC00:34
dansmithwhat's your point00:34
*** dpawlik has quit IRC00:34
dansmith?00:34
artomErr, good question :)00:34
*** zhubx has joined #openstack-nova00:34
dansmithit still requires a non-trivial environment to test for real, as does multiple cells00:35
artomI guess you only need 1 (or 2, for migrations) NUMA machines00:35
artomFor real multi-cell you'd need... how much, actually?00:35
dansmithyou can do it all in one box, but it's not trivial to set up00:35
dansmitheither way,00:35
dansmithyou're drawing parallels that don't matter... it's a complicated thing that is hard to test for real, so extra work in making more-than-unit tests are worthwhile.. It all depends on some unfortunately turbo-global state, which make those hard00:36
dansmithnone of that points to "meh, it's hard, so just punt on tests" to me00:36
artomThat wasn't really the thinking00:37
artomIf func tests are super hard, I'd rather invest time and effort into setting up a real CI00:38
dansmithI was going on this: [16:27:04]  <artom>Reason I'm asking is - without this, I don't see a way to func test NUMA LM meaningfully00:38
dansmithand00:38
dansmith[16:30:21]  <artom>dansmith, so, we can follow up on ML, but I think that rules out NUMA LM func testing?00:38
artomdansmith, more in the sense of "I won't bother attempting it"00:40
artomI spoke to the openlab guys a bit this morning00:40
artomMy request seemed to make sense for them00:40
artomSo I'd rather push on that00:40
artomAnd if that turns out to be impossible, we can revisit func tests00:41
*** itlinux has quit IRC00:44
alex_xumriedem_away: thanks, let me go through the last around today00:52
*** dpawlik has joined #openstack-nova01:02
*** whoami-rajat has joined #openstack-nova01:04
*** _fragatina has quit IRC01:04
*** dpawlik has quit IRC01:07
*** tbachman has joined #openstack-nova01:08
*** itlinux has joined #openstack-nova01:09
*** slaweq has joined #openstack-nova01:11
*** hamzy has joined #openstack-nova01:11
*** itlinux has quit IRC01:11
*** itlinux has joined #openstack-nova01:12
*** slaweq has quit IRC01:16
*** ttsiouts has quit IRC01:17
*** itlinux has quit IRC01:18
*** itlinux has joined #openstack-nova01:19
*** wolverineav has quit IRC01:21
Kevin_Zhengdansmith: OK, I will check it now01:21
*** wolverineav has joined #openstack-nova01:26
*** wolverineav has quit IRC01:26
*** wolverineav has joined #openstack-nova01:27
*** itlinux has quit IRC01:51
*** itlinux has joined #openstack-nova01:52
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (3)  https://review.openstack.org/57410401:53
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (4)  https://review.openstack.org/57410601:54
*** wolverineav has quit IRC01:54
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (5)  https://review.openstack.org/57411001:54
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (6)  https://review.openstack.org/57411301:54
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (7)  https://review.openstack.org/57497401:54
*** wolverineav has joined #openstack-nova01:57
*** wolverineav has quit IRC01:57
*** wolverineav has joined #openstack-nova01:57
*** mlavalle has quit IRC01:58
*** itlinux has quit IRC01:58
*** tkajinam has quit IRC02:02
*** tkajinam has joined #openstack-nova02:03
*** tkajinam has quit IRC02:07
*** tkajinam has joined #openstack-nova02:10
*** Dinesh_Bhor has joined #openstack-nova02:21
*** markvoelker has quit IRC02:26
*** mriedem_away has quit IRC02:40
*** wolverineav has quit IRC02:40
openstackgerritMatt Riedemann proposed openstack/nova master: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree  https://review.openstack.org/64018102:40
openstackgerritMatt Riedemann proposed openstack/nova master: Make nova-grenade-live-migration voting and gating  https://review.openstack.org/64018202:40
*** gyee has quit IRC02:42
*** hongbin has joined #openstack-nova02:46
*** sdake has joined #openstack-nova02:54
*** dpawlik has joined #openstack-nova03:03
*** awalende has joined #openstack-nova03:06
*** dpawlik has quit IRC03:08
*** awalende has quit IRC03:10
*** slaweq has joined #openstack-nova03:11
*** psachin has joined #openstack-nova03:12
*** slaweq has quit IRC03:15
*** zhubx has quit IRC03:17
*** zhubx has joined #openstack-nova03:17
*** dave-mccowan has joined #openstack-nova03:18
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (8)  https://review.openstack.org/57531103:24
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (9)  https://review.openstack.org/57558103:24
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (10)  https://review.openstack.org/57601703:24
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (11)  https://review.openstack.org/57601803:25
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (12)  https://review.openstack.org/57601903:25
*** harlowja has quit IRC03:27
*** markvoelker has joined #openstack-nova03:27
*** zhubx has quit IRC03:35
*** dpawlik has joined #openstack-nova03:35
*** zhubx has joined #openstack-nova03:35
*** awaugama has quit IRC03:36
*** janki has joined #openstack-nova03:39
*** dpawlik has quit IRC03:40
*** zhubx has quit IRC03:54
*** zhubx has joined #openstack-nova03:55
*** bhagyashris has joined #openstack-nova03:57
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (13)  https://review.openstack.org/57602004:00
*** markvoelker has quit IRC04:00
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (14)  https://review.openstack.org/57602704:00
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (15)  https://review.openstack.org/57603104:01
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (16)  https://review.openstack.org/57629904:01
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (17)  https://review.openstack.org/57634404:02
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (18)  https://review.openstack.org/57667304:02
*** dave-mccowan has quit IRC04:04
*** sridharg has joined #openstack-nova04:06
*** sdake has quit IRC04:08
*** slaweq has joined #openstack-nova04:11
*** wolverineav has joined #openstack-nova04:11
*** slaweq has quit IRC04:16
*** wolverineav has quit IRC04:17
*** Dinesh_Bhor has quit IRC04:29
*** hongbin has quit IRC04:40
*** wolverineav has joined #openstack-nova04:42
*** Dinesh_Bhor has joined #openstack-nova04:50
*** tetsuro has joined #openstack-nova04:53
openstackgerritsean mooney proposed openstack/nova master: libvirt: delegate ovs plug to os-vif  https://review.openstack.org/60243204:55
*** markvoelker has joined #openstack-nova04:57
yonglihemriedem: thanks.05:24
*** zhubx has quit IRC05:29
*** zhubx has joined #openstack-nova05:30
*** markvoelker has quit IRC05:31
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (19)  https://review.openstack.org/57667605:32
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (20)  https://review.openstack.org/57668905:32
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (21)  https://review.openstack.org/57670905:32
openstackgerritTakashi NATSUME proposed openstack/nova master: Remove mox in unit/network/test_neutronv2.py (22)  https://review.openstack.org/57671205:33
*** dpawlik has joined #openstack-nova05:36
*** dpawlik has quit IRC05:41
*** wxy-xiyuan has quit IRC05:41
*** wolverineav has quit IRC05:43
*** ratailor has joined #openstack-nova05:44
*** igordc has quit IRC05:56
*** Dinesh_Bhor has quit IRC05:59
*** Dinesh_Bhor has joined #openstack-nova06:02
*** mtreinish has quit IRC06:04
*** wolverineav has joined #openstack-nova06:05
openstackgerritYongli He proposed openstack/nova master: Add server sub-resource topology API  https://review.openstack.org/62147606:09
*** slaweq has joined #openstack-nova06:11
*** dpawlik has joined #openstack-nova06:12
*** slaweq has quit IRC06:15
*** dpawlik has quit IRC06:17
*** tetsuro has quit IRC06:20
*** markvoelker has joined #openstack-nova06:28
*** itlinux has joined #openstack-nova06:33
*** ivve has joined #openstack-nova06:34
*** itlinux_ has joined #openstack-nova06:37
*** itlinux has quit IRC06:40
*** zhubx has quit IRC06:42
*** zhubx has joined #openstack-nova06:42
*** Luzi has joined #openstack-nova06:43
*** zhubx has quit IRC06:46
*** lbragstad has quit IRC06:49
*** mtreinish has joined #openstack-nova06:50
*** mdbooth_ has joined #openstack-nova06:53
openstackgerritTakashi NATSUME proposed openstack/python-novaclient master: Remove unnecessary if statement  https://review.openstack.org/64026606:54
*** mdbooth has quit IRC06:57
*** wolverineav has quit IRC06:57
*** markvoelker has quit IRC07:00
*** slaweq has joined #openstack-nova07:11
openstackgerritChris Friesen proposed openstack/nova master: Add emulated TPM support to Nova  https://review.openstack.org/63136307:12
openstackgerritChris Friesen proposed openstack/nova master: Add support for resize and cold migration of emulated TPM files  https://review.openstack.org/63993407:12
*** slaweq has quit IRC07:15
*** dpawlik has joined #openstack-nova07:15
*** wolverineav has joined #openstack-nova07:25
*** tetsuro has joined #openstack-nova07:28
*** slaweq has joined #openstack-nova07:32
*** takashin has left #openstack-nova07:36
*** itlinux_ has quit IRC07:37
*** itlinux has joined #openstack-nova07:39
*** sdake has joined #openstack-nova07:44
*** itlinux has quit IRC07:55
*** markvoelker has joined #openstack-nova07:57
*** Dinesh_Bhor has quit IRC08:01
*** tetsuro has quit IRC08:01
*** Dinesh_Bhor has joined #openstack-nova08:08
*** wolverineav has quit IRC08:10
*** awalende has joined #openstack-nova08:11
*** tkajinam has quit IRC08:18
*** ccamacho has joined #openstack-nova08:19
*** sdake has quit IRC08:23
*** tssurya has joined #openstack-nova08:25
openstackgerritBoxiang Zhu proposed openstack/nova master: Raise error when force_raw_images and images_type are incompatible  https://review.openstack.org/64027108:26
*** tesseract has joined #openstack-nova08:31
*** markvoelker has quit IRC08:31
kashyapgibi: Morning, when you get a quick minute: on something 'cosmetic': do you think my comment makes sense here: https://review.openstack.org/#/c/637834/4/nova/conf/libvirt.py08:32
*** luksky has joined #openstack-nova08:34
*** mcgiggler has joined #openstack-nova08:39
*** tosky has joined #openstack-nova08:42
openstackgerritMerged openstack/nova master: Adds the server group info into show server detail API.  https://review.openstack.org/62147408:45
tssuryayonglihe: ^^ finally :)08:47
*** sdake has joined #openstack-nova08:55
*** ttsiouts has joined #openstack-nova08:56
*** yan0s has joined #openstack-nova08:58
*** ralonsoh has joined #openstack-nova09:02
gibikashyap: I agree with you and alex_xu on that review.09:03
*** rambo_li has joined #openstack-nova09:04
kashyapgibi: Also, afraid that patch is wrong on several levels09:04
kashyapI'll review and comment with suggestions09:04
gibikashyap: thanks for helping that patch09:04
*** panda|ruck|off is now known as panda|ruck09:05
kashyapgibi: Also, related is this important (IMHO) blueprint that needs to be addressed: https://blueprints.launchpad.net/nova/+spec/cpu-selection-with-hypervisor-consideration09:05
kashyapI added it to the Train Etherpad.09:06
gibikashyap: is it something that is available for us as we bumped min libvirt and qemu versions?09:07
gibistephenfin: you might missed my ping yestarday about https://review.openstack.org/#/c/625311/25/doc/source/admin/networking.rst@266 and I'll be on and off this morning but I'd like to make progress on that patch09:07
kashyapgibi: Afraid, no.  Those two newer APIs need newer version constants.09:08
kashyapgibi: When you get a moment, can you tell if that requires a specification?09:08
kashyapMaybe I'll just write a short one, it helps to write out background, motivation and why this matters in a spec09:08
gibikashyap: does it affect any config option or nova API?09:09
gibias far as I understand it is a libvirt driver only change09:09
kashyapYes09:09
kashyapgibi: No, doesn't affect config options.  At its core: it rips out two older, ineffective APIs, in Nova09:09
kashyapAnd adds in 'better' and sensible equivalent replacements09:09
*** IvensZambrano has joined #openstack-nova09:10
gibiif there is no config or REST API change then I think we don't need a spec. Still if you think that a sort spec would help then go ahead with that09:10
gibithe more interesting thing is that we cannot rip out the old code before we bump the min versions again09:11
kashyapYep, noted.  Doesn't affect REST API.  Nor config.09:11
*** priteau has joined #openstack-nova09:11
openstackgerritMerged openstack/nova master: Add remove_resources_from_instance_allocation to report client  https://review.openstack.org/63965309:13
*** rambo_li has quit IRC09:14
*** wolverineav has joined #openstack-nova09:17
*** rambo_li has joined #openstack-nova09:17
*** rambo_li has quit IRC09:21
*** wolverineav has quit IRC09:22
*** markvoelker has joined #openstack-nova09:28
*** rambo_li has joined #openstack-nova09:28
*** mdbooth_ is now known as mdbooth09:35
mdboothcfriesen: Apologies for being obtusely British ;) Seems you managed to decode it.09:35
*** dtantsur|afk is now known as dtantsur09:37
*** derekh has joined #openstack-nova09:42
*** rambo_li has quit IRC09:43
*** bhagyashris has quit IRC09:49
openstackgerritZhenyu Zheng proposed openstack/nova master: Add compute service support for attach/detach root volume  https://review.openstack.org/61475009:54
openstackgerritZhenyu Zheng proposed openstack/nova master: Detach/Attach root volume API changes  https://review.openstack.org/62398109:54
openstackgerritBalazs Gibizer proposed openstack/nova master: Add pf_interface_name tag to passthrough_whitelist  https://review.openstack.org/62531109:56
openstackgerritBalazs Gibizer proposed openstack/nova master: Add pf_interface_name tag to passthrough_whitelist  https://review.openstack.org/62531109:57
*** markvoelker has quit IRC10:01
*** ttx has quit IRC10:07
*** ttx has joined #openstack-nova10:08
openstackgerritSurya Seetharaman proposed openstack/nova master: Add "links" in the response of GET /servers/{server_id} for a down-cell instance  https://review.openstack.org/64030210:10
*** cdent has joined #openstack-nova10:15
openstackgerritSurya Seetharaman proposed openstack/nova master: Add "links" in the response of "nova show" for a down-cell instance  https://review.openstack.org/64030210:15
*** owalsh has quit IRC10:17
*** sdake has quit IRC10:20
*** eandersson has quit IRC10:20
*** rambo_li has joined #openstack-nova10:22
*** owalsh has joined #openstack-nova10:31
openstackgerritTheodoros Tsioutsias proposed openstack/nova master: Add requested_networks to RequestSpec  https://review.openstack.org/57020110:34
openstackgerritTheodoros Tsioutsias proposed openstack/nova master: Enable rebuild for instances in cell0  https://review.openstack.org/57020310:34
openstackgerritTheodoros Tsioutsias proposed openstack/nova master: Introduce the PENDING instance state  https://review.openstack.org/56647310:34
openstackgerritTheodoros Tsioutsias proposed openstack/nova master: Allow rebuild for instances in PENDING state  https://review.openstack.org/63758510:34
openstackgerritZhenyu Zheng proposed openstack/nova master: Detach/Attach root volume API changes  https://review.openstack.org/62398110:35
*** rambo_li has quit IRC10:49
*** erlon has joined #openstack-nova10:55
*** luksky has quit IRC10:56
*** markvoelker has joined #openstack-nova10:58
openstackgerritMatthew Booth proposed openstack/nova master: Improve libvirt image and snapshot handling  https://review.openstack.org/61669211:01
*** IvensZambrano has quit IRC11:05
*** luksky has joined #openstack-nova11:09
*** IvensZambrano has joined #openstack-nova11:11
*** cdent has quit IRC11:12
*** ttsiouts has quit IRC11:18
*** cdent has joined #openstack-nova11:20
stephenfinmdbooth: Could you test something real quick for me?11:21
stephenfin$ tox -e functional -- -n nova/tests/functional/libvirt/test_pci_sriov_servers.py11:21
stephenfinDoes that fail? ^11:22
*** luksky has quit IRC11:22
mdboothstephenfin: Sec11:23
mdboothstephenfin: Do you mean to have .py on the end, there?11:23
mdboothSeems like that wouldn't match anything?11:23
stephenfinmdbooth: Never mind. I used '--recreate' and it now works. I was seeing privsep warnings previously11:23
stephenfinmdbooth: Yeah, I'm using the '-n' flag11:23
stephenfinLet's me give a path to a file instead of a module path. Makes test runs waaay faster11:24
*** Dinesh_Bhor has quit IRC11:24
mdboothNot familiar with that11:24
mdboothNice11:24
mdboothSounds super useful11:24
mdboothstephenfin: Also lets you find import ordering bugs :)11:24
stephenfinYou can also do nova/tests/functional/libvirt/test_pci_sriov_servers.py::SRIOVServersTest.test_create_server_with_PF11:24
stephenfinIndeed11:24
*** IvensZambrano is now known as snevi11:24
*** markvoelker has quit IRC11:30
yonglihetssurya: thanks. gonna to rebase another one.11:31
*** amrith has joined #openstack-nova11:33
*** amrith has quit IRC11:35
*** amrith has joined #openstack-nova11:36
sean-k-mooneystephenfin: why not just use -- SRIOVServersTest.test_create_server_with_PF11:39
stephenfinsean-k-mooney: You could, but if there are overlapping test names, that lets you be more specific11:40
stephenfinwithout needing to convert the entire thing to a Python module path11:40
sean-k-mooneywll you cant have overlaping test names in the same module11:40
stephenfinnope, but you can between different modules11:40
stephenfinof different classes, if you didn't specify that11:41
openstackgerritYongli He proposed openstack/nova master: Add server sub-resource topology API  https://review.openstack.org/62147611:41
mdboothAlso you lose the speed advantage, because stestr still has to load everything and then filter for your test11:41
mdboothLoading everything is what takes such a long time11:41
sean-k-mooneyyes SRIOVServersTest is ca class and test_create_server_with_PF is a test function you can add one level o module to make it unique11:41
sean-k-mooneyanyway -n is still useful11:41
sean-k-mooneyso is --failing11:42
stephenfinI was literally just about to point that out :)11:42
mdboothAvoiding loading everything is why I have my trivial 'testone' bash alias11:42
mdboothI can run a single test in a couple of seconds, rather than waiting 20 seconds for stestr to load all tests before filtering to find my 1 test11:42
*** erlon_ has joined #openstack-nova11:43
sean-k-mooneymdbooth: not that long in the grand schme of things a full unit test run takes 8 minues event with the dest discovery running a monudes tests is less then a minute much less typically11:43
mdboothBut super useful when you're iterating on 1 test.11:44
sean-k-mooneymdbooth: ya if i use vscode to click on a test it runs instantly11:44
sean-k-mooneyand i can debug it :)11:44
*** amrith has quit IRC11:44
stephenfinsean-k-mooney: Sounds like a blog :)11:45
sean-k-mooneybut i typically do most of my deve with nano or emacs and run the test by had out of habit11:46
sean-k-mooneystephenfin: how about i raise you os-vif docs https://review.openstack.org/#/c/638058/4/doc/source/devref/vs-code.rst11:46
*** erlon has quit IRC11:46
sean-k-mooneyi need to test that htis works properly for nova and ill write up somthing similar. i have mainly been using it on os-vif lately to debug the functional tests11:47
*** Luzi has quit IRC11:50
*** ttsiouts has joined #openstack-nova11:51
*** jchhatbar has joined #openstack-nova12:06
*** janki has quit IRC12:07
*** yaawang has quit IRC12:10
*** yaawang has joined #openstack-nova12:13
*** panda|ruck is now known as panda|ruck|lunch12:16
*** priteau has quit IRC12:17
*** luksky has joined #openstack-nova12:18
*** threestrands has quit IRC12:24
aspiershi folks, I found a bug in assertXMLEqual12:27
*** markvoelker has joined #openstack-nova12:28
aspiersI'm fixing it ... hopefully it won't show up other bugs in our tests12:28
sean-k-mooneyoh ? what was broken12:30
aspierssean-k-mooney: hard to explain briefly but will make sense when you see the review12:40
aspiers(hopefully ;-)12:40
aspiersof course I meant assertXmlEqual() - typo12:41
*** gibi is now known as giblet12:45
*** amrith__ has joined #openstack-nova12:45
*** amrith__ has left #openstack-nova12:46
aspiersI wonder why we invented our own xml differ12:53
aspiershttps://xmldiff.readthedocs.io/en/stable/api.html#main-diffing-api looks good to me12:54
cdent<sparta>12:54
*** wolverineav has joined #openstack-nova12:54
aspiersmaybe xmldiff didn't exist in 2012 when Kevin Mitchell wrote ours12:55
sean-k-mooneyaspiers: https://pypi.org/project/xmldiff/#history12:55
sean-k-mooneyit did but did not have its 1.0.0 release until 201812:56
sean-k-mooneyit also did not have any release between 2005 and 201512:56
sean-k-mooneyso it had been unmaintianed for 7 years at that point12:57
aspiersahah12:57
aspiersStill a generic xml differ never really belonged in nova. I'm sure other projects could use one12:58
sean-k-mooneyah it also looks like they did a complete rewrite form c to python between 0.6 and 1.0 so it an entirely different code base12:58
aspiersYes12:58
openstackgerritSurya Seetharaman proposed openstack/nova master: Add "links" in the response of GET /servers/{server_id} for a down-cell instance  https://review.openstack.org/64030212:58
aspiersMaybe we should switch to that in the future12:58
*** wolverineav has quit IRC12:59
tssuryastephenfin: I fixed the nit in https://review.openstack.org/640302, but the +2 went off :( ,12:59
tssuryathanks for the review13:00
sean-k-mooneyaspiers: yes perhaps. am i the only one who gets suspices of project with 100% test coverage13:00
aspiershaha no you are not ;-)13:00
sean-k-mooneyfor some reason it always make me trust them less13:00
aspiersxmldiff has been around since 2001 https://github.com/Shoobx/xmldiff/blob/272300bb29b8c2d444971e8051aae7bf513cdb98/ChangeLog13:00
aspiersalthough as you say it might not have been very palatable for many years13:00
*** markvoelker has quit IRC13:01
sean-k-mooneyya in anycase if its maintained and it works we could use it13:01
sean-k-mooneythat said i dont think the code is overly big ofr assertXmlEquals either13:01
openstackgerritSurya Seetharaman proposed openstack/nova master: Add "links" in the response of "nova show" for a down-cell instance  https://review.openstack.org/64030213:02
openstackgerritBalazs Gibizer proposed openstack/nova master: Ensure that bandwidth and VF are from the same PF  https://review.openstack.org/62354313:03
*** panda|ruck|lunch is now known as panda|ruck13:04
openstackgerritBalazs Gibizer proposed openstack/nova master: Ensure that bandwidth and VF are from the same PF  https://review.openstack.org/62354313:04
*** ratailor has quit IRC13:07
openstackgerritBalazs Gibizer proposed openstack/nova master: Support server create with ports having resource request  https://review.openstack.org/63636013:11
gibletjaypipes: Hi! I've replied in https://review.openstack.org/#/c/625310/31/nova/objects/instance_pci_requests.py@39 If you still think we should rename then let me know and I will jump on it13:13
jrollefried: alex_xu: this was rebased, want to re-approve? https://review.openstack.org/#/c/636326/13:15
*** mdbooth_ has joined #openstack-nova13:23
*** mdbooth has quit IRC13:24
bauzasefried: alex_xu: jaypipes: others, I finally removed my -W because I'm not sure I can provide the fix for the problem I saw by this week13:30
bauzasinstead, I'll provide a documentation13:30
*** mriedem has joined #openstack-nova13:32
*** eharney has quit IRC13:33
efriedjroll: done13:35
efriedbauzas: ack. My +2 is still on there... for now :P13:36
jrollthanks13:36
bauzasefried: heh thanks13:36
efriedjroll: you knew there was no way you were going to merge a patch at PS1.13:36
jrollefried: so close though13:37
*** tbachman has quit IRC13:44
*** marst has joined #openstack-nova13:45
*** efried is now known as fried_rice13:45
yonglihehope this would in reviewing queue -:) https://review.openstack.org/#/c/621476/13:46
openstackgerritMerged openstack/nova master: fix up numa-topology live migration hypervisor check  https://review.openstack.org/63535013:46
mriedemyonglihe: are you working on adding the related novaclient change for 2.70?13:47
mriedem*2.71 i mean13:47
*** jmlowe has quit IRC13:48
yongliheyeah, on my queue.13:50
openstackgerritMatt Riedemann proposed openstack/nova stable/rocky: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree  https://review.openstack.org/64019113:51
yongliheby default , it could show up. i need to dig how to good fit to nova client, might be some documentation work i guss.13:51
yongliheguess13:51
mriedemit should just work, but you have to at least bump the supported api version and it would be good to modify the fakes response in unit tests to show it in the output for nova show/update/rebuild13:52
mriedemso this https://github.com/openstack/python-novaclient/blob/master/novaclient/__init__.py#L2813:52
yonglihesure, got it.13:52
mriedemthe fake responses for unit tests are here https://github.com/openstack/python-novaclient/blob/master/novaclient/tests/unit/v2/fakes.py13:53
yonglihenice, thanks.13:53
mriedemso something like this https://github.com/openstack/python-novaclient/blob/master/novaclient/tests/unit/v2/fakes.py#L64713:53
tssuryamriedem, melwitt, dansmith: I have this very easy doc change on the archive nova-manage command just to clarify what max-rows means there: https://review.openstack.org/#/c/606995/. Please take a look at it when you have time,13:53
mriedemyonglihe: you can check the requested microversion from the fake method and return the server_groups key in the response13:54
*** marst has quit IRC13:54
mriedemlike this https://github.com/openstack/python-novaclient/blob/master/novaclient/tests/unit/v2/fakes.py#L202313:54
yonglihemriedem:  ok, thanks.13:56
openstackgerritMatt Riedemann proposed openstack/nova master: Update --max-rows parameter description for archive_deleted_rows  https://review.openstack.org/60699513:57
mriedemtssurya: done13:58
*** markvoelker has joined #openstack-nova13:58
*** mlavalle has joined #openstack-nova14:00
*** sdake has joined #openstack-nova14:01
*** _mlavalle_1 has joined #openstack-nova14:01
*** mlavalle has quit IRC14:04
*** Sundar has joined #openstack-nova14:08
*** nicholas has joined #openstack-nova14:09
fried_riceHey folks, especially folks like sean-k-mooney: when we do a resize-to-same-host, do neutron ports get unbound and rebound or do we leave them alone?14:14
* fried_rice hopes the latter14:14
*** TheJulia is now known as needssleep14:14
fried_riceSundar: ^14:15
ttsioutsmriedem: I tried to address your comments in https://review.openstack.org/#/c/570203/. would be great to hear your feedback when you have the time. thanks in advance14:16
*** mcgiggler has quit IRC14:17
*** snevi has quit IRC14:17
*** snevi has joined #openstack-nova14:17
sean-k-mooneyfried_rice: i would have to check they proable are not but it might be resonable to require that they are14:17
sean-k-mooneyi dont htink we allow changing vif_model via flavor im pretty sure that is image only14:17
fried_ricesean-k-mooney: I'm thinking it could be problematic for bandwidth resources, if there are paths in which we would end up binding to different providers somehow - the allocations would wind up screwy. giblet, have you already considered this?14:19
*** jchhatbar has quit IRC14:19
sean-k-mooneyam if the numa toplogy changes as part of a resize then in the future the bandwith allcoation might need to be updated14:19
*** sdake has quit IRC14:19
fried_ricesean-k-mooney: And in anticipation of cyborg, similarly, if we had to unbind and rebind accelerators.14:19
*** eharney has joined #openstack-nova14:19
sean-k-mooneyfried_rice: well we hit the schudler as part of a resize so if the flavor chagned the device-profile for example in the cyborge case we defiently would need to rebind the device14:20
gibletfried_rice: do you mean changing the flavor might affect the port allocation?14:20
fried_ricegiblet: actually I'm more concerned about the case where changing the flavor does/should *not* affect the port allocation/binding.14:21
sean-k-mooneyit may14:21
*** jmlowe has joined #openstack-nova14:21
fried_ricelike, if all I do is increase VCPU, is it possible that I wind up with my vifs being swizzled?14:21
Sundarsean-k-mooney: What about rebuild? It could be on same host or different host. In the former case, AFAIK, it is only a detach but not an unbind.14:21
sean-k-mooneySundar: rebuild is confied to the same host14:22
sean-k-mooneySundar: it will never do a migrate14:22
*** sdake has joined #openstack-nova14:22
Sundarsean-k-mooney: If the rebuild changes the image, and the new image has different traits, is that not a problem?14:23
sean-k-mooneyfried_rice: i think there are edgecses such as changin numa toplogy that might require it if you have sriov port or hardwar offload ovs or you use the numa aware vswitch feature14:23
sean-k-mooneySundar: we will fail the rebuild if its not valid for the current host14:23
mriedemrebuild doesn't mess with port bindings14:23
sean-k-mooneyrebuilt with same image bypasses the scheduler14:23
sean-k-mooneyrebuilt with different hits the schduler14:23
gibletdoes resize-to-same host is just allows but never forces the resize to keep the server on the same host14:23
mriedemresize to same host calls the same methods to change port bindings, but i think the neutron api code we have doesn't do the update if the host doesn't change14:24
mriedemhttps://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L323014:24
*** sdake has quit IRC14:24
* mordred thanks the nova team for having local hacking check infrastructure in place he could steal from14:24
sean-k-mooneygiblet: right if the resize was invalid on the same host it would do a cold migraton to a vliad host if one exits in that case14:25
sean-k-mooneymriedem: in princiable a rebind with a new image could contain tratis that may affect the neutro ports but in practice we have never need to cater for that case todate so ya for rebuild we have never needed to touch the port bindings14:27
gibletfried_rice: to answer your original question server move operations (even resize on same host) are not covered by the current implementation so your question needs to be answered during Train14:27
fried_ricegiblet: Okay, good to know.14:27
gibletfried_rice: my general idea regarding move is, to re-query resoruce request from neutron ports, include it to the scheduling. conductor will transparently handle migration allocation for ports then.14:28
fried_ricegiblet: It would be nice to prefer candidates where the ports don't move.14:28
fried_ricewhich sounds like more hacked up matching.14:29
gibletfried_rice: so during resize-to-same-host case it would be nice to keep the ports allocated to from the same RP, I seenow14:29
sean-k-mooneyfried_rice: for what it worth the image trait edgecase for neutron port i am tink of would be a a required or forbding trati for a partacalr nic vendor when using sriov.  we talked about this as a way for image to expres i have drivers for vendor X but not Y14:29
gibletfried_rice: the problem with that is nova-scheduler always selects the first allocation for a host even if there is more than one candidate is returned for that host14:30
gibletby placement14:30
sean-k-mooneyfried_rice: well its not really more hacked up match and more a weigher.14:31
*** markvoelker has quit IRC14:31
fried_riceweighers are by definition hacks14:31
fried_rice...said someone I'm sure14:31
sean-k-mooneyha perhaps14:32
gibletfried_rice: https://github.com/openstack/nova/blob/68d4cb44edfe7b5f85618254cc906dcef44f7ce3/nova/scheduler/filter_scheduler.py#L23814:32
fried_riceyup14:33
sean-k-mooneybut unlike filter they are intended to handel the optimisation of the best host rather then filter which is what we want in thiscase. although as giblet mething we might need to look at multiple allcoation candiates which the weigher dont play a role in14:33
*** ccamacho has quit IRC14:33
*** ccamacho has joined #openstack-nova14:34
gibletnova filter scheduler does not filter and weigh allocation candidates today14:34
gibletit filter and weigh hosts only14:34
sean-k-mooney yes im aware14:34
sean-k-mooneybut i think what fried_rice was asserting sounded like a weigher of the allocation candiates14:34
gibletso we cannot prefer the candidate that keeps the port allocated from the same RP during resize-to-same-host14:35
fried_ricesean-k-mooney: right: prefer candidates with [these ports]14:35
fried_ricegiblet: yes.14:35
fried_ricewe can't, but we should.14:35
sean-k-mooneycan you use in_tree to help14:35
*** lbragstad has joined #openstack-nova14:35
gibletfried_rice: soo we have to extend the filter interface to know about allocation candidates not just hosts14:35
sean-k-mooneyi guess that would force the same host14:35
fried_ricewell, we should be using in_tree for resize-to-same-host anyway.14:35
fried_riceBut that's unrelated to the problem of not moving the ports.14:36
fried_ricegiblet: I was just talking to mriedem about that the other day.14:36
fried_riceand Sundar as well.14:36
sean-k-mooneyfried_rice: were we not talking about a subtree version too  i dont know if we could restict the neutron request to the same RPs they had previously14:36
fried_ricesean-k-mooney: Yes, there's a spec up for that.14:36
sean-k-mooneyi dont think the placement sysntax is that granuarly14:36
gibletfried_rice: can we tell our preference to placement about return those a_c first that keeps the port allocated from the same RP?14:37
fried_ricenope, not today.14:37
fried_ricegiblet: Not yet, but with the subtree thing we might be able to.14:37
sean-k-mooneyya so resize to same host would basice add in_tree to the numbered requst group for the prot resouce or whatever the spec says14:37
sean-k-mooneyfried_rice: sound like a thing to solve in train14:38
fried_riceRe allocation candidates to filters: I was thinking we ought to extend BaseFliter with a new method that accepts the allocation candidate under consideration, as well as the (relevant subset of the) provider_summaries. We do the try/except NotImplemented and fall back to the old method so we don't break existing filters. But new filters can then be written to take that information into account.14:38
sean-k-mooneyfried_rice: ya that came up before14:39
gibletfried_rice: hm, do you have a link to subtree a_c spec?14:39
fried_ricesec14:39
*** sdake has joined #openstack-nova14:39
sean-k-mooneyfried_rice: i think passing the allcoation candiates to the filters/werigher makes sence.14:40
fried_ricegiblet: Hah, it's yours: https://review.openstack.org/#/c/595236/14:40
gibletfried_rice: but that is subtree for GET /resource_providers not subtree for a_c14:41
fried_ricemm, well, in the spirit of in_tree, we should implement it for GET /a_c as well.14:41
fried_ricegiblet: tetsuro has started code: https://review.openstack.org/#/c/636092/ (but for GET /rps)14:42
sean-k-mooneyfried_rice: i think that proably should be implemented at the resouce group level not the top level14:42
gibletfried_rice: I haven't tought about that direction yet. The use case in my spec is purely to limit the amount of data neutron needs to read from placemente14:42
fried_ricemriedem: Currently (merged code) does resize-to-same-host even use GET /a_c at all?14:42
*** _mlavalle_1 has quit IRC14:42
sean-k-mooneye.g. this requst_group must be form this sub tree14:43
*** mlavalle has joined #openstack-nova14:43
sean-k-mooneythat would allow us to say the general request must be form the compute node sub tree but the neutron bandwith must be form the subtree of the ortinginal RP that the allcoation were form14:43
fried_ricesean-k-mooney: just so. and/or should be from this exact provider. The latter would be better for the case we're talking about.14:44
sean-k-mooneyyes the formor would allow nesting to work better when we model sriov nics in placmeent14:44
gibletwait, we don't want to force resize-to-same-host to only work for the same host. We still want to get back candidates for other hosts. So we cannot simply restrict the a_c querty to a given tree14:44
gibletor subtree14:45
fried_riceeh? By definition, resize-to-same-host is... to the same host. Isn't it??14:45
gibletfried_rice: nope14:45
sean-k-mooneyboth the bandwith request and vf request woudl be in the same group but they could be in different provdiers depending on how we model it14:45
gibletfried_rice: it is a config option that allows resize to use the same host too14:45
*** bnemec is now known as beekneemech14:46
sean-k-mooneyfried_rice: it does not gurarentee it14:46
fried_riceI think mriedem is working on making it so? https://review.openstack.org/#/c/638791/14:46
gibletfried_rice: allow_resize_to_same_host14:46
sean-k-mooneyfried_rice: by default we dont allow resize to same host. when you enable it in the nova.conf it just allows it as a candiate but unlees you also force the host it could be resized to a different host14:47
sean-k-mooneyfried_rice: mriedem is working on removing the double allocation problem in the case we actully do resize on the same host14:48
fried_ricemm14:48
fried_ricemriedem: In fact, do we even use GET /a_c for resize in general, or do we just build the new allocs by parsing the new flavor??14:49
gibletthere is a full scheduling for resize including a_c query14:50
*** marst has joined #openstack-nova14:50
gibletonly rebuild has a partial scheduling (no a_c)14:50
gibletafaik14:50
*** artom is now known as temka14:50
*** ccamacho has quit IRC14:51
gibletfried_rice: resize is just a cold_migrate with a changing flavor https://github.com/openstack/nova/blob/68d4cb44edfe7b5f85618254cc906dcef44f7ce3/nova/conductor/tasks/migrate.py#L13414:53
*** ccamacho has joined #openstack-nova14:53
fried_ricehm14:53
*** snevi has quit IRC14:54
*** snevi has joined #openstack-nova14:54
gibletfried_rice: here is the select_destination call https://github.com/openstack/nova/blob/68d4cb44edfe7b5f85618254cc906dcef44f7ce3/nova/conductor/tasks/migrate.py#L23514:54
*** awaugama has joined #openstack-nova14:54
fried_riceokay.14:55
*** awalende has quit IRC15:00
mriedemfried_rice: yes resize to same host does GET /a_c15:00
*** awalende has joined #openstack-nova15:00
fried_riceokay.15:00
fried_ricemriedem: Then something I just sent on the ML may be entirely wrong.15:01
*** jaypipes is now known as leakypipes15:01
fried_riceo/ leakypipes15:01
*** tbachman has joined #openstack-nova15:01
*** itlinux has joined #openstack-nova15:01
tssuryamriedem: thanks!15:02
cfriesenmdbooth_: no worries.  thanks for fixing it up.15:03
*** tbachman has quit IRC15:03
*** Sundar is now known as DarthNau15:03
*** tobias-urdin is now known as tobeass-urdin15:03
*** awalende has quit IRC15:04
*** tbachman has joined #openstack-nova15:07
*** lbragstad is now known as elbragstad15:08
cfriesenmdbooth: I assume that change has your approval since you resubmitted it?15:09
*** sdake has quit IRC15:10
*** itlinux has quit IRC15:12
melwitto/15:12
fried_ricejroll: Thanks for that email. Turns out this will affect (at least) nova-powervm's OOT driver https://github.com/openstack/nova-powervm/blob/master/nova_powervm/virt/powervm/driver.py#L23015:13
fried_riceo/ melwitt15:13
*** wolverineav has joined #openstack-nova15:15
melwittfried_rice: we were considering adding a warning log to the patch if a driver needs to be updated. do you think that would be useful or do you think the reno alone is good enough?15:15
fried_ricemelwitt: I think it's a good idea.15:15
fried_ricesince we're going to the trouble of being nice about changing the interface.15:16
*** wolverineav has quit IRC15:20
kashyapmriedem: Thanks for bringing this to my attention: https://review.openstack.org/#/c/637834/ (Select cpu model from a list of cpu models)15:20
kashyapmriedem: Reviewed it this morning.  And it needs a lot more rework15:20
kashyapIt is doing several things that are terribly ineffective, even poorly duplicating libvirt's work15:20
kashyapWonder if you (dis)agree with my remarks there15:22
*** ratailor has joined #openstack-nova15:23
*** dpawlik has quit IRC15:24
*** ivve has quit IRC15:26
*** markvoelker has joined #openstack-nova15:28
*** itlinux has joined #openstack-nova15:29
*** igordc has joined #openstack-nova15:30
mriedemkashyap: i'll defer to you15:30
mriedemi haven't been reviewing the change15:30
mriedembigger fires to fight15:31
mriedemjroll: so i guess you're going to add a log message?15:31
mriedemfor the TypeError?15:31
kashyapmriedem: I think all the points I (and Alex Xu) raised should be addressed.15:32
kashyapI'll follow through15:32
*** itlinux has quit IRC15:36
*** chason_ has quit IRC15:36
*** itlinux has joined #openstack-nova15:38
*** jistr is now known as jistr|mtg15:38
*** belmoreira has quit IRC15:39
*** chason has joined #openstack-nova15:41
openstackgerritMatt Riedemann proposed openstack/nova master: Stop using "nova" in API samples when creating a server  https://review.openstack.org/63987415:43
openstackgerritBen Nemec proposed openstack/nova master: Add oslo.privsep to config-generator list  https://review.openstack.org/64040415:45
mriedembauzas: so what changed with the reboot issue you were holding https://review.openstack.org/#/c/599208/ up for?15:46
*** chason has quit IRC15:46
*** betherly has joined #openstack-nova15:48
openstackgerritLee Yarwood proposed openstack/nova master: Use migration_status during volume migrating and retyping  https://review.openstack.org/63722415:49
*** temka has quit IRC15:49
openstackgerritMerged openstack/nova master: Remove port allocation during detach  https://review.openstack.org/62242115:50
openstackgerritBalazs Gibizer proposed openstack/nova master: Add pf_interface_name tag to passthrough_whitelist  https://review.openstack.org/62531115:51
openstackgerritBalazs Gibizer proposed openstack/nova master: Ensure that bandwidth and VF are from the same PF  https://review.openstack.org/62354315:51
openstackgerritBalazs Gibizer proposed openstack/nova master: Support server create with ports having resource request  https://review.openstack.org/63636015:51
*** chason has joined #openstack-nova15:52
*** artom has joined #openstack-nova15:53
gibletstephenfin: added an extra sentence about grouping multiple PF in a single expression in https://review.openstack.org/#/c/62531115:53
*** artom is now known as temka15:53
*** betherly has quit IRC15:53
stephenfingiblet: Sounds good. I'll get to it before EOD. Been working on adding functional tests for https://review.openstack.org/#/c/624444/215:53
*** ratailor has quit IRC15:54
gibletstephenfin: thanks. hit me with the link of the functional test and I will review it15:54
sean-k-mooneydo you think we will land that in stein by the way. it would be good to backport if we do15:55
jrollfried_rice: mriedem: yeah, I'll add it shortly here15:55
jrollthanks15:55
openstackgerritStephen Finucane proposed openstack/nova master: objects: Store InstancePCIRequest.numa_policy in DB  https://review.openstack.org/62444415:55
openstackgerritStephen Finucane proposed openstack/nova master: WIP: fakelibvirt: Add ability to generate fake PCI devices  https://review.openstack.org/64040915:55
fried_riceight15:56
gibletsean-k-mooney: the actual fix looked good to me so I can help pushing it through15:56
stephenfingiblet: There you go. I was trying to fix some TODOs in the tests but that can be done some other time https://review.openstack.org/62444415:56
gibletstephenfin: looking15:56
*** sdake has joined #openstack-nova15:58
sean-k-mooneyon a slightly unrelated note i would like to land this bugfix in stien too if we can https://review.openstack.org/#/c/602432/7 but i need to do some more testing of it locally. i think it is correct but i want to be certin before we merge it and say the bugs are closed.15:59
*** tssurya has quit IRC15:59
*** yan0s has quit IRC15:59
openstackgerritAdam Spiers proposed openstack/nova master: fix bug with XML matcher handling missing children  https://review.openstack.org/64041116:01
*** markvoelker has quit IRC16:01
*** jistr|mtg is now known as jistr16:02
sean-k-mooneyactully stephenfin giblet should i add a release note for that16:02
aspierssean-k-mooney: ^^^16:02
stephenfinsean-k-mooney: For?16:02
*** sdake has quit IRC16:03
stephenfinsean-k-mooney: Can you take a look at https://review.openstack.org/#/c/625311/25/doc/source/admin/networking.rst ?16:05
stephenfinsean-k-mooney: Just because I know you've expressed concerns about using device names as identifiers in the past16:05
sean-k-mooneystephenfin: https://review.openstack.org/#/c/602432/7 since it close https://bugs.launchpad.net/neutron/+bug/1734320 and https://bugs.launchpad.net/neutron/+bug/181598916:06
openstackLaunchpad bug 1734320 in OpenStack Compute (nova) "Eavesdropping private traffic" [Undecided,In progress] - Assigned to sean mooney (sean-k-mooney)16:06
openstackLaunchpad bug 1815989 in neutron "OVS drops RARP packets by QEMU upon live-migration causes up to 40s ping pause in Rocky" [Undecided,In progress] - Assigned to sean mooney (sean-k-mooney)16:06
stephenfinsean-k-mooney: though I am imagining the ship has sailed based on giblet's comments about neutron using them for RP IDs16:06
sean-k-mooneysure ill take a look now16:06
*** sdake has joined #openstack-nova16:06
stephenfinsean-k-mooney: I love release notes and think we should have them for all (user-facing) bugs. However, there's a user visible change being introduced here so I'd definitely say add one, yeah16:08
sean-k-mooneystephenfin: if the pf is directly assignied to a guest and then readdd to the host its may or may not have a netdev associated with it and it may our may not have the same name16:08
stephenfinSo16:08
sean-k-mooneyso the sriov agent already depend on the pci address in the port profile so why not use that instad of a dev name16:09
gibletsean-k-mooney: the sriov agent is configured with device names in neutron16:10
sean-k-mooneyi have not read the full spec16:10
lyarwoodmdbooth_: https://review.openstack.org/#/q/topic:bug/1803961+(status:open+OR+status:merged) - would you mind clearing your reviews out of that topic and hitting mine again early next week?16:10
sean-k-mooneygiblet: yes and it looks up the pci address16:10
stephenfinCould we add code that says if 'pf_interface_name': 'eth0' is set, that same device can't be included whitelisted itself?16:10
sean-k-mooneygiblet: nova passes only the pci address to neutron in the port profile for the vf16:10
gibletsean-k-mooney: yes, that is true16:11
gibletsean-k-mooney: I need the neutron devs of this feature to argue why the device name was choosen for the RP nameing16:12
gibletstephenfin: you want some kind of sanity check on the configured value to see if there are contradictions?16:12
*** macza has joined #openstack-nova16:12
*** macza has quit IRC16:12
sean-k-mooneyprobaly becasue fo the connetx-3 nic form mellononx which has two pfs on teh same pciaddress16:13
stephenfingiblet: I'm not sure yet. I'm trying to decide if it's a good idea or not :)16:13
*** macza has joined #openstack-nova16:13
sean-k-mooneybut that didnt work with openstack until verry verry recently and has been supperseeded16:13
*** mdbooth_ is now known as mdbooth16:13
sean-k-mooneymaybe it was connectx-216:13
mdboothlyarwood: phase 1 complete16:13
*** wwriverrat has quit IRC16:14
sean-k-mooneygiblet: there may also be confusion between teh existing "devname" field and teh new "pf_interface_name"16:14
gibletstephenfin: if the check only depends on the actula configured value and does not need to call out to the virt driver or other 3rd parties then I can be convinced to add a check16:14
lyarwoodmdbooth: ack thanks16:14
gibletsean-k-mooney: jeah the existing devname is a filtering field while the new pf_interface_name is a tag16:15
cfriesencan anyone figure out why this is failing in zuul?  http://logs.openstack.org/63/631363/14/check/openstack-tox-py27/f6aa9a3/testr_results.html.gz   It works fine for me locally.16:15
sean-k-mooneygiblet: ya didnt we say we were not going to add more tag to the whitelist like 3 or 4 cycles ago16:15
gibletsean-k-mooney: if there is devname provided in the filter nova could copy that as a pf_inteface_name tag automatically16:15
gibletsean-k-mooney: I did not know about that rule. what should I use instead to drive the pci claim?16:16
stephenfingiblet: The devname would only match a single PF or VF though16:16
sean-k-mooneygiblet: well we can also lookup the pf name via sysfs16:16
gibletsean-k-mooney: if there is more devname for a PF then we don't know which one to use16:16
stephenfinsean-k-mooney: Yeah, but as giblet pointed out, there can be a mismatch for those ConnectX-3 NICs16:16
sean-k-mooneyyes that is true but do we want to cater for that edgecase16:17
stephenfinbauzas: Before you knock off for the weekend, fancy hitting https://review.openstack.org/#/c/626932/ up?16:17
stephenfinsean-k-mooney: Aye, that was going to be my question16:17
bauzasstephenfin: looking16:17
sean-k-mooneywe could but it feels like we are workign around a silicon/driver bug16:17
bauzasah this one16:17
stephenfinyuuup.16:17
bauzas+d16:17
sean-k-mooneygiblet: stephenfin by the way only the PF had a duplicate pci adress16:18
gibletsean-k-mooney: if nova looks up the name form sysfs and then selects one then the neutron admin needs to use the same one in the config, but we cannot really tell the admin which one nova selected16:18
sean-k-mooneythe vf could still be assocated with the pf netdev via sysfs16:19
sean-k-mooneygiblet: the admin already has to set up the config correctly on both the nova and neturon side16:19
stephenfinLet's double check this16:20
gibletsean-k-mooney: no not really as  so far if the neutron admin selected any of the devname it worked as the ip tool accepted any of the devname. But now the name became a key between the nova and neutron communication to decide which PF the VF is allocated from16:20
sean-k-mooneyif we add the pf_interface_name all pci whitelist will have to be updated to make the bandwith based schduling work16:20
*** luksky has quit IRC16:20
sean-k-mooneygiblet: but its not a key we send today and modifying the whitelist iis an upgrade impact16:21
gibletsean-k-mooney: yes, as well as every neutron config needs to be updated to enable bandwidth16:21
sean-k-mooneywhat we use as a key today is the pci address16:21
stephenfinsean-k-mooney: I doubt there's much that can be done about that at this point though16:22
*** kmalloc is now known as needscoffee16:22
sean-k-mooneywhat you are proposing would work but it feels like a hack that we should be able to avoid without modifing the config16:22
gibletsean-k-mooney: if the admin does not add the bandwith inventory to the neutron agent config then she dont need to modify the whitelist either16:22
openstackgerritChris Friesen proposed openstack/nova master: Add emulated TPM support to Nova  https://review.openstack.org/63136316:22
openstackgerritChris Friesen proposed openstack/nova master: Add support for resize and cold migration of emulated TPM files  https://review.openstack.org/63993416:22
sean-k-mooneygiblet: yes but this can be done without modifying the whitelist at all16:22
stephenfinsean-k-mooney: Let's just make sure it can though16:23
stephenfinWe can't access sysfs outside of a driver, right?16:23
*** sdake has quit IRC16:23
gibletsean-k-mooney: yes, if neutron changes the naming to the RPs to use pci address for example16:23
sean-k-mooneystephenfin: the driver could add the pf_interface_tag itself if we really wanted it too automatically16:24
gibletsean-k-mooney: and we change pci claim to allow selecting PF for a PCIrequest by PF pci address16:24
sean-k-mooneyso neutron should not force nova to select a pf16:24
sean-k-mooneyit does not know the numa affintity so it cant make the right choice16:25
gibletsean-k-mooney: it is nova and placement that forces the pf selection16:25
gibletsean-k-mooney: port has resource request that is fulfilled by a placemetn allocation candidate that will become an allocation16:25
gibletsean-k-mooney: and that forcing the PF16:25
gibletsean-k-mooney: because the bandwidth is allocated from an RP that representes a PF16:26
sean-k-mooneythe port cant force the PF16:26
sean-k-mooneyit can ask for bandwith but its nova/placemtn aht has to match the port and bandwith up16:26
gibletsean-k-mooney: if the bandwidth is represented in placement then nova, by selectin an a_c, actually forces a PF16:26
sean-k-mooneyyes it will...16:27
sean-k-mooneysice we have a bug in our pci affinity policies it wont break existing workloads16:28
gibletso when the allocation is done in placement nova needs to drive the pci claim mechanism to select VF from the same PF as the bandwidth is allocated from16:28
sean-k-mooneyyes i know i raised that point a few tiems when i argured we could not remove the reosuce tracker16:29
gibletone way to drive it is to know which RP was used for the bandwidth, get the name of it, and from the name learn the pf device name16:29
sean-k-mooneygiblet: so the request in the neutron port does not contain a devename right16:30
gibletsean-k-mooney: right.16:30
sean-k-mooneyok so its placemetn that choses the bandwtih resouce provider and we are just corralting it16:30
gibletsean-k-mooney: yes16:31
gibletsean-k-mooney: we have to make sure the pci claim chooses the same PF as placement16:31
sean-k-mooneyi would still prefer if we looked up the pf name automatically in the driver and populated the tag rather then making it a manual step in the whitelist16:32
stephenfinsean-k-mooney: Out of curiosity, doesn't systemd solve the "attach to guest, detach, and still have the same name" issue? https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/16:32
*** sapd1 has quit IRC16:32
stephenfinsean-k-mooney, giblet: I must say, so would I16:32
sean-k-mooneyno not in all cases16:32
gibletsean-k-mooney: for that nova needs to select the same device name as the neutron admin selected in the neutron config16:33
stephenfinGiven that it's information that's very easy to extract, and the only reason I see not to do that is because of a weird bug in an effectively EOL card16:33
sean-k-mooneyif the pf is not rebound to the nic driver after its bound to vfio-pci it will not have a netdev anymore16:33
sean-k-mooneygiblet: yes that is not that hard to do16:34
gibletsean-k-mooney, stephenfin: is it doable in 4 effective day before FF?16:34
sean-k-mooneywe have similar code to do that in os-vif already https://github.com/openstack/os-vif/blob/master/vif_plug_ovs/linux_net.py#L313-L33416:35
cfriesencould I get some eyes on https://review.openstack.org/#/c/620706/ ?  It's on a runway.16:36
sean-k-mooneygiblet: we need to look up netdev names form pci address to make hardware offload ovs work16:36
*** psachin has quit IRC16:38
sean-k-mooneygiblet: you jsut do open( "/sys/bus/pci/devices/%s/physfn/net" % (pci_addr)).readline().strip()16:38
gibletsean-k-mooney: is it fine to do that during the startup of the nova-compute service while parsing the passthrough_whitelist?16:39
sean-k-mooneyi dont know. we do it every time we plug a hardwar offloaed ovs-port and we plug all ports on startup16:39
sean-k-mooneygranted its os-vif that executing it but os-vif is run as part of the nova comptue agent so it shoudl be fine16:40
stephenfingiblet: I'm guessing we could just slot it in here https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L5926-L594016:41
gibletstephenfin, sean-k-mooney: thanks for the pointers I have to sit down with a clear mind and play with that codepath to see how can I inject the extra info there in a way that it will be used by the pci claim16:44
sean-k-mooney giblet we also have a call to lookup device by name just after that16:44
sean-k-mooneyhttps://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L595816:44
sean-k-mooneyso doing the invers should be fine16:44
openstackgerritBen Nemec proposed openstack/nova master: Add oslo.privsep to config-generator list  https://review.openstack.org/64040416:44
sean-k-mooneygiblet: im not really objecting to useing a tag by the way16:45
sean-k-mooneygiblet: i think we can just automate it16:46
stephenfingiblet: Sounds good. If you can't make progress fairly quickly, I'd be willing to hold my nose so we could get it in. Would be good to try avoid it though, given the few issues there are already16:46
gibletsean-k-mooney: I think I understand. I agree in general to automate it if possible.16:46
gibletsean-k-mooney: I'm more affraid that I will not be able to make it work16:47
*** DarthNau has quit IRC16:47
gibletstephenfin: alternatively I can remove this patch from the series, declare the two PF on the same compute on the same physnet as not-supported in Stein16:47
stephenfine.g. confusion with 'devname', issues if you whitelist more than PF and need for extra (possibly unnecessary) configuration, the first two of which probably need additional code to fix16:47
stephenfingiblet: Aye, I was wondering if it was possible to shove this to the end and de-risk it. If that's an option, it might be a good one16:48
sean-k-mooneygiblet: i think that is what we did for vgpus16:48
sean-k-mooneyits not ideal but it would work. i would be good to ask leakypipes or dansmith16:49
*** sridharg has quit IRC16:49
sean-k-mooneythey tend to have stong opipions on this topic16:49
*** itlinux has quit IRC16:49
gibletstephenfin: yeah, I think the sortness of time will push me to remove this patch from the critical path16:50
sean-k-mooneyactully maybe you shouldnt ask them :)16:50
*** luksky has joined #openstack-nova16:51
sean-k-mooneygiblet: i can try and hack something up at the weekend to do auto tagging and then you can decide on monday16:51
*** Sundar has joined #openstack-nova16:51
gibletsean-k-mooney: if you hack someting up then I will definitly look at it on Monday but please don't spend your free time on this crazy things :)16:52
sean-k-mooneyi spent my free time last weekend trying to autoamte a tox nutika enviorment ot automaticlaly compile nova and then execute the unit test.16:54
sean-k-mooneythis might be a less crazy16:54
gibletsean-k-mooney: :)16:54
sean-k-mooneyi got it to work for os-vif but nova make it explode16:54
leakypipessean-k-mooney, stephenfin: ask me what?16:55
stephenfinleakypipes: We're not overly happy about this patch and think pf_interface_name could be figured out dynamically by the driver https://review.openstack.org/#/c/625311/2916:56
sean-k-mooneynoting you will like. unless your opipion on adding tags to the pci whitelist has changed recently16:57
stephenfinleakypipes: giblet is concerned about time constraints though and think maybe declaring that two PF on the same compute on the same physnet as not-supported in Stein would be a safer option16:57
stephenfin*maybe declaring two PFs16:58
*** jangutter has quit IRC16:58
leakypipesstephenfin: ok, that patch is near the top of my queue. will try to comment on it this afternoon.16:58
*** markvoelker has joined #openstack-nova16:58
gibletI'm not too affraid of saying that this edge case is not supported in Stein16:59
sean-k-mooneywe are already doing auto discovery of nic offloads exctra on startup and we have code to discover the name of pf form vfs in os-vif so i was suggesing just looking up the pf name on startup as part of the pci device discovery16:59
giblethowever I will only be able to declare this case unsupported in the documentation as I don't see an easy way to detect the case from code and reject or just fail the boot request17:01
gibletso if the admin ignores the doc then she can create a situation when the bandwidth is allocated from a PF while the VF is allocated from other PF for the same port17:02
stephenfingiblet: Is there a file missing from https://review.openstack.org/#/c/623543/ ?17:04
stephenfinI'm trying to figure out where "Then the pci claim will enforce that the PF interface name in the request matches the interface name of the PF from where the VF is selected from." happens17:04
stephenfin(from the commit message, line 53)17:04
*** gyee has joined #openstack-nova17:04
gibletstephenfin: I think the pci claim matches the request with the device spec field by field17:05
stephenfinOh, wait, tags17:05
gibletso if the pf_interface_name is in the request as well as in the device spec then it will be matched17:05
stephenfinyup. duh17:05
stephenfinIt's clearly Friday :)17:05
gibletsoo Friday17:05
sean-k-mooneyhttps://www.youtube.com/watch?v=kfVsfOSbJY017:06
*** ttsiouts has quit IRC17:06
sean-k-mooneyits stuck in my head so i give it to you too17:07
* stephenfin gives sean-k-mooney the evil eyes (virtually)17:07
*** ivve has joined #openstack-nova17:08
gibletit is heavy stuff :)17:11
*** snevi has quit IRC17:18
openstackgerritJim Rollenhagen proposed openstack/nova master: ironic: check fresh data when sync_power_state doesn't line up  https://review.openstack.org/63669917:20
openstackgerritJim Rollenhagen proposed openstack/nova master: Remove TypeError handling for get_info  https://review.openstack.org/64004317:20
*** panda|ruck is now known as panda|ruck|off17:21
* giblet goes away for the weekend17:22
openstackgerritMatt Riedemann proposed openstack/nova master: Check hosts have no instances for AZ rename  https://review.openstack.org/50920617:25
mriedembauzas: i'm good with ^ again17:28
mriedemwe should all thank avolkov for his persistence17:28
*** cdent has quit IRC17:28
*** markvoelker has quit IRC17:31
*** betherly has joined #openstack-nova17:40
openstackgerritMatt Riedemann proposed openstack/nova master: api-ref: explain aggregate set_metadata semantics  https://review.openstack.org/64046017:41
mriedemstephenfin: cfriesen: can one of you backport https://review.openstack.org/#/c/635350/ to stable/rocky?17:44
cfriesenmriedem: can do17:44
openstackgerritStephen Finucane proposed openstack/nova stable/rocky: fix up numa-topology live migration hypervisor check  https://review.openstack.org/64046217:45
cfriesenokay17:45
cfriesen:)17:45
stephenfinno merge conflicts :)17:45
temkastephenfin, hey, sean-k-mooney's saying on the mailing list that you somehow managed to get different configs to different services in func tests. Any recollection of that?17:46
stephenfinummm...17:46
temka(No link to ML post because the web UI is apparently lagging behind)17:47
*** sdake has joined #openstack-nova17:47
temkastephenfin, I'm normally artom, btw, in case you're wondering17:47
stephenfintemka: I'm _guessing_ he's referring to 45662d77a2da77714f8e792e86ebd64a52270ef517:49
*** betherly has quit IRC17:50
stephenfintemka: I don't think it's possible to have different nova.conf configurations in functional tests though17:50
temka(Okay, turns out this friday nick idea wasn't necessarily the best, because 'temka' is normally what my mom calls me)17:51
stephenfinthat config is global so as soon as you'd change it, every service would see the change17:51
temkastephenfin, that's what I thought17:51
* temka proceeds with his original plan17:52
aspiersmelwitt, mriedem: if we are retargeting SEV for Train, what is the correct approach to updating the spec/bp?17:54
*** betherly has joined #openstack-nova17:54
mriedemhttps://specs.openstack.org/openstack/nova-specs/readme.html#previously-approved-specifications17:55
aspiershah, sorry - still asking newbie FAQs :)17:55
aspiersbreton: ^^^17:55
*** mriedem is now known as mriedem_lunch17:56
*** dtantsur is now known as dtantsur|afk18:00
*** mvkr has quit IRC18:02
*** jaosorior has quit IRC18:04
*** mvkr has joined #openstack-nova18:06
*** IvensZambrano has joined #openstack-nova18:07
*** betherly has quit IRC18:07
*** eharney has quit IRC18:09
*** irclogbot_2 has joined #openstack-nova18:09
*** wolverineav has joined #openstack-nova18:10
*** wolverineav has quit IRC18:10
*** wolverineav has joined #openstack-nova18:10
*** betherly has joined #openstack-nova18:11
*** tesseract has quit IRC18:13
*** IvensZambrano has quit IRC18:13
*** betherly has quit IRC18:15
*** betherly has joined #openstack-nova18:17
openstackgerritAdam Spiers proposed openstack/nova master: fix bug with XML matcher handling missing children  https://review.openstack.org/64041118:20
*** derekh has quit IRC18:22
*** ralonsoh has quit IRC18:22
*** wolverineav has quit IRC18:23
*** wolverineav has joined #openstack-nova18:25
*** mrhillsman is now known as mrhillsman_lunch18:28
*** markvoelker has joined #openstack-nova18:28
openstackgerritAdam Spiers proposed openstack/nova master: Parse <emulator> elements from virConnectGetCapabilities()  https://review.openstack.org/64048318:28
*** betherly has quit IRC18:30
*** betherly has joined #openstack-nova18:37
*** ivve has quit IRC18:39
*** betherly has quit IRC18:42
*** betherly has joined #openstack-nova18:45
*** wolverineav has quit IRC18:50
*** wolverineav has joined #openstack-nova18:53
*** mdbooth_ has joined #openstack-nova18:55
*** betherly has quit IRC18:57
*** cdent has joined #openstack-nova18:58
*** mdbooth has quit IRC18:58
*** cdent has quit IRC18:58
*** sdake has quit IRC19:00
*** betherly has joined #openstack-nova19:01
*** markvoelker has quit IRC19:01
*** awalende has joined #openstack-nova19:03
*** ircuser-1 has quit IRC19:03
*** ircuser-1 has joined #openstack-nova19:05
*** awalende_ has joined #openstack-nova19:05
openstackgerritAdam Spiers proposed openstack/nova master: Add detection of SEV support from QEMU/AMD-SP/libvirt on AMD hosts  https://review.openstack.org/63385519:06
*** betherly has quit IRC19:07
*** awalende has quit IRC19:08
*** wolverineav has quit IRC19:12
openstackgerritAdam Spiers proposed openstack/nova master: Add detection of SEV support from QEMU/AMD-SP/libvirt on AMD hosts  https://review.openstack.org/63385519:15
openstackgerritAdam Spiers proposed openstack/nova master: Add new "supports_amd_sev" capability to libvirt driver  https://review.openstack.org/63868019:17
*** purplerbot has quit IRC19:18
*** purplerbot has joined #openstack-nova19:18
*** jdillaman has quit IRC19:18
*** betherly has joined #openstack-nova19:26
openstackgerritMerged openstack/python-novaclient master: Remove unnecessary if statement  https://review.openstack.org/64026619:27
temkaaspiers, btw, going by your commit message in https://review.openstack.org/#/c/638680/4, you know you can submit multiple patches as a stack/series for review, right?19:27
aspierstemka: that's what I've done :)19:28
*** mriedem_lunch is now known as mriedem19:28
temkaaspiers, ok cool - you were calling out a dependency needlessly in your commit message, so I wanted to double check19:29
aspierstemka: well, I don't think it's entirely needless. I think it helps explain the dependency19:29
temkaaspiers, as you wish :)19:29
*** mrhillsman_lunch is now known as mrhillsman19:29
aspiersI'm a believer in "too much info is better than not enough" - at least in this context19:29
temkaI'm not sure I'd agree, actually :)19:30
aspiersnot in general19:30
temkaHeh, not even in this context19:30
aspiersIt's easier for reviewers to skip over stuff which is obvious to them than for them to be forced to figure out non-obvious stuff19:30
*** wolverineav has joined #openstack-nova19:30
aspiersThe former is quicker than the latter19:30
*** betherly has quit IRC19:30
aspiersI'm generalizing, of course :)19:30
temkaFor me at least, it implies that there's a dependencies that's not already apparent in the patch series19:31
aspiersbut I think it's true in this particular scenario19:31
temkaSo I went looking, and realized you were just making it really explicit19:31
aspiersThe thing is, a patch series is linear, but a dependency tree is a DAG19:31
aspiersThe serialization is lossy, so it is not trivial to infer the tree from the series19:32
temkaHeh19:32
aspiersCoincidentally, I've done a lot of work on this topic outside OpenStack: https://github.com/aspiers/git-deps19:32
aspiersBut anyway, I welcome all nitpicks, so thanks ;-)19:33
temkaaspiers, heh, I can't bring myself to concentrate fully on what I'm trying to do, so instead I nitpick others's work :/19:33
aspiersLOL I have exactly the same problem ;-)19:34
temkaI had no idea of your level of experience with Gerrit, so just wanted to make sure you're not pushing one patch at a time and manually specified the deps19:34
aspierstemka: Yeah thanks, it's definitely worth checking that with unfamiliar faces^Wnicks :-)19:35
aspiersI'm pretty familiar with Gerrit. Attended the last User Summit which was super fun and interesting.19:35
*** eharney has joined #openstack-nova19:35
temkaBut then you started talking about DAGs and I realized I'm *probably* barking up the wrong tree ;)19:35
aspiers;)19:36
temkaMind you, I've seen people graduate with CS degress and not be able to write a single line of C, so :P19:36
aspiersThis is true19:36
aspiersgit and its ecosystem is actually one of my specialist areas19:36
aspiersI've hacked on git a bit, and built a plethora of tools. Since it's Friday, you can find more ramblings at https://blog.adamspiers.org/tag/git/ if you're curious19:38
temka"Structured Procrastination"19:38
temkaSounds about right19:38
aspiersAlso will finish adding https://ethercalc.openstack.org/github-gerrit to the Contributor's Guide one of these days ... maybe :-o19:39
*** betherly has joined #openstack-nova19:39
temkaaspiers, wanna add https://github.com/berrange/gerrymander in the Gerrit clients square?19:41
aspiersooh, that's a new one for me I think19:41
aspierstemka: please feel free19:41
temkaNot sure it's still maintained, it's been a few years since last commit19:42
*** tosky has quit IRC19:42
*** tosky has joined #openstack-nova19:43
aspiersadd it anyway19:43
temkaDone19:43
aspiersmriedem: SEV trait successfully detected on real hardware using capabilities :) https://pasteboard.co/I3rPhvJ.png19:44
*** ivve has joined #openstack-nova19:47
*** betherly has quit IRC19:48
*** irclogbot_2 has quit IRC19:49
mriedemget right out of town19:52
aspiers:)19:52
aspiersplus all those new capability traits too of course19:53
*** wolverineav has quit IRC19:56
mriedemaspiers: so you've already worked around the virConnectGetDomainCapabilities thing?19:57
cfriesensean-k-mooney: ran into a problem with the cold migration with TPM.../var/lib/libvirt/swtpm and the directories under it are owned by root and nova-compute is not running as root.19:57
*** wolverineav has joined #openstack-nova19:57
aspiersmriedem: yes, https://review.openstack.org/#/c/633855/7/nova/virt/libvirt/host.py@71119:57
*** markvoelker has joined #openstack-nova19:58
aspiersmriedem: for SEV, this is *almost* good enough for now; I think the only possible issue is that if testing on another distro, qemu-kvm might live in a different place19:58
aspiersI don't have a RHEL / Fedora box handy to check19:59
aspiersmriedem: and the code in comments is pretty darn close to what it will look like when that patch does land20:00
*** irclogbot_2 has joined #openstack-nova20:01
*** wolverineav has quit IRC20:02
*** Sundar has quit IRC20:13
*** betherly has joined #openstack-nova20:16
*** sdake has joined #openstack-nova20:18
*** sdake has quit IRC20:18
temkaDammit, even something as simple as 2 different hostnames for the 2 compute services is going to be annoying20:19
temkahttps://github.com/openstack/nova/blob/master/nova/tests/unit/virt/libvirt/fakelibvirt.py#L113120:20
mriedemmelwitt: i'm confused about your comment here https://review.openstack.org/#/c/633350/7/nova/objects/instance_mapping.py@4220:20
*** betherly has quit IRC20:21
mriedemtemka: ah that's where that comes from...i was wondering because that question came up in a functional test that gibi wrote which was using the fakelibvirt stuff20:21
mriedemcan't you modify Connection to pass in a hostname which defaults to 'compute1'?20:21
mriedemnote that self.start_service() in the tests will allow you to specify the binary (compute) and hostname which is what our multi-host functional tests do,20:22
mriedembut those aren't using the fakelibvirt stuff20:22
*** sdake has joined #openstack-nova20:22
temkamriedem, it's solved in the fake virt driver, but that's different because it's a virt drive20:23
temkafakelibvirt isn't a driver - the real libvirt is the driver20:24
temkafakelibvirt just mocks out the parts that call the libvirt API20:24
mriedemi.e. https://github.com/openstack/nova/blob/master/nova/virt/fake.py#L65420:24
temkamriedem, right, and virt/fake.py replaces virt/libvirt/driver.py20:25
mriedemyeah i know20:25
temkafakelibvirt.py replaces what virt/libvirt/driver.py calls down to20:25
mriedemright the python-libvirt interface20:25
temkaRight20:25
*** derekh has joined #openstack-nova20:26
mriedemdansmith: am i correct in my understanding of what you're asking for here? https://review.openstack.org/#/c/633350/7/nova/objects/instance_mapping.py@4220:27
mriedemon load of instance mapping, of it doesn't have a user_id, try to get it from the request spec, failing that, try the build request, failing *that* try the instance via the cell mapping20:27
mriedems/of/if/20:27
*** liuyulong has quit IRC20:27
dansmithmriedem: right, the nullability has nothing to do with where or how we get the user_id to fill it in20:28
*** wolverineav has joined #openstack-nova20:28
temkaI guess I have to mock ComputeManager._get_nodename()20:29
temka... per service20:29
mriedemdansmith: which means we wouldn't need to update the instance mapping on read of the instance from the api https://review.openstack.org/#/c/638574/4/nova/compute/api.py@240420:29
mriedemalthough, if the request spec doesn't have user_id, and the build request is gone, and the cell the instance is in is down...20:29
mriedemthen i guess you're sol?20:30
mriedemtemka: you can't override the fakelibvirt.Connection hostname value it uses?20:30
aspiersOK, weekend time. Have a good one everyone o/20:30
*** markvoelker has quit IRC20:31
temkamriedem, I guess I could, but I can't make it different depending on which compute service is calling it20:31
dansmithtemka: mock service.manager._get_nodename() after it's started to return the value it needs to keep returning, instead of mocking the class20:32
temkadansmith, tried that just now, didn't work20:33
temkaI think it needs to be mocked *before* (somehow) it's started20:33
dansmithmriedem: yeah, I mean, making that value nullable for the down-cell-and-not-anywhere-else case is not reasonable I think20:33
temkaBecause upon startup it reports to placement, and there's a hostname conflict20:33
dansmithtemka: you have to mock it both places,20:33
*** wolverineav has quit IRC20:33
*** betherly has joined #openstack-nova20:33
*** phasespace has joined #openstack-nova20:33
dansmithtemka: which is why I said "needs to keep returning"20:34
temkadansmith, which both places?20:34
dansmiththe class will catch the startup case, but be invalid once you move to the next service, so you need to mock the resulting instance of the manager class to keep returning the value ongoing20:34
temkaAh, I see20:34
dansmithwe used to be able to pass in a different manager class to start_service i think, which would let you just subclass per node.. can we still do that?20:35
temkaSo mock fakelibvirt's getHostname, start compute0, mock its get_nodename, mock getHostname to the 2nd thing, start compute1, mock its get_nodename()20:35
temkaThat feels... racy20:35
temkaActually no20:36
dansmithtemka: there are no threads here so, no20:36
mriedemheh you still can technically classload a compute manager class it looks like20:37
dansmithmriedem: yeah that's what I was thinking of20:37
*** betherly has quit IRC20:37
mriedemself.start_service('compute', host='fake-host1', manager='artom.special.fake.ComputeManager')20:38
temkaIf anyone says 'dependency injection' there's going to be violence20:38
dansmithyeah20:38
temkaI *still* have no idea what that is20:38
*** betherly has joined #openstack-nova20:38
temkaDespite university's best attempt to drill it into our heads with massive bloated Java frameworks20:39
mriedemi should have gone to university to learn about hospital20:39
temkadansmith, woot, that worked20:39
mriedemcolour flavour20:39
mriedemdansmith: so my one reservation on loading the instance mapping user_id from the turducken is that down cell case because i'm not sure what we'd do, likely 500 from the api, but you can't really online data migrate your way out of that b/c the online data migration relies on, you guessed it, the cells being up20:40
dansmithmriedem: yeah20:41
dansmithyou're also migrating from a state where cells had to be up,20:41
*** jmlowe has quit IRC20:41
dansmithso until all that data is patched up, it's not surprising that...all cells have to be up20:41
*** sdake has quit IRC20:42
openstackgerritMerged openstack/nova master: Ironic: bump minimum API version to 1.38  https://review.openstack.org/63632620:44
*** betherly has quit IRC20:47
*** igordc has quit IRC20:48
*** igordc has joined #openstack-nova20:49
*** wolverineav has joined #openstack-nova20:51
temka"Allocation for resource provider 'e7fbbec5-7f95-4f4e-895c-a14b691bca6e' that does not exist"20:56
temkaLooks like I'm missing a placement-y fixture or something20:56
mnaserdoes nova cache neutron port info somewhere?20:57
mriedemmnaser: instance_info_caches table20:57
mnaseri had a misconfigured n-ovs-agent which lead to resizes failing with binding_failed on the port20:57
mnaseri've reset the info, but its still grabbing the cached value20:58
mnaserok, i'll dig there, thanks mriedem20:58
mriedemmnaser: see https://review.openstack.org/#/c/591607/20:58
mriedemthe instance_info_caches.network_info is a json blob string so if you're looking to mess with that manually....20:59
mnasermriedem: neat. i assume its not a real cache as in i cant really just drop that row20:59
mriedemummm20:59
mriedemi don't think that will work21:01
mriedembecause anything that tries to refresh the cache from neutron will just rely on the existing cache, which will now be empty...21:02
mriedemwhich is what https://review.openstack.org/#/c/591607/ was trying to address21:02
*** tbachman has quit IRC21:03
mriedemwholesale deleting of the cache would also make nova forget which ports nova created and which were pre-created, so on server delete nova will also delete any pre-existing ports21:03
mriedemwhich is probably not something your users want21:03
mriedemrelated change here https://review.openstack.org/#/c/603844/21:04
mriedemthe issue is kind of this code right here https://github.com/openstack/nova/blob/c57d532a306303412d6253f424bee7d665865279/nova/network/neutronv2/api.py#L305521:05
mriedemif you wipe out that record in the db, that will result in port_ids being empty21:05
mriedemso https://github.com/openstack/nova/blob/c57d532a306303412d6253f424bee7d665865279/nova/network/neutronv2/api.py#L3061 won't do anything21:05
mriedemit's kind of dumb - if port_ids was empty but current_neutron_port_map is not, we should probably realize "oh we're fully rebuilding the cache b/c the cache was wiped out"21:06
mnasermriedem: yeah so cache is probably not really cache i guess21:07
mnaserit is tracking state21:07
mriedemwell it's a cache in the sense that you don't have to go to neutron every time you want to find out what ports are attached to a server,21:07
mriedemsimilar to the block_device_mappings table is a "cache"21:07
mriedemof volumes attached to the server21:07
mnaseryeah21:08
mriedembut yeah it's got some state information in there too about the ports nova knows it created21:08
*** erlon_ has quit IRC21:08
mnaser(ps i totally thought this whole detach => delete port thing is silly but thats another topic)21:08
mnaserthis whole thing would be so much easier if i just detached the port and attached it again21:09
mnaserbut because it was done on boot, detaching the port = deleting it21:09
* mnaser thinks an explicit detach should always detach not "delete if it was created by nova"21:09
mriedemoh there was a spec for that...21:09
mriedemi.e. delete_on_termination for ports21:09
mriedemlike volumes21:09
*** betherly has joined #openstack-nova21:11
mriedemanyway like i said if you delete the cache via the db, and port_ids is empty but current_neutron_port_map is not https://github.com/openstack/nova/blob/c57d532a306303412d6253f424bee7d665865279/nova/network/neutronv2/api.py#L3061 you could probably just force a cache refresh21:11
cfriesencan someone take a look at http://logs.openstack.org/63/631363/15/check/openstack-tox-py27/6b108d3/testr_results.html.gz and suggest why test_executable_exists_in_path is failing?  It works perfectly on my local box.21:13
openstackgerritMatt Riedemann proposed openstack/nova master: WIP: Force refresh nw info cache if cache is empty but ports attached  https://review.openstack.org/64051621:16
mriedemmnaser: like this ^21:16
*** awalende has joined #openstack-nova21:21
*** awalende_ has quit IRC21:21
*** betherly has quit IRC21:23
*** tbachman has joined #openstack-nova21:24
mriedemdansmith: here is another fun side effect of the mox->mock test removal - besides making backports a nightmare, it also makes git blame a real chore21:27
mriedemthat is all21:27
*** betherly has joined #openstack-nova21:27
dansmithyeah21:27
mriedemplus most parts of nova you can't git blame in github anymore b/c it times out21:27
dansmith...really?21:28
*** markvoelker has joined #openstack-nova21:28
dansmithwith git or github?21:28
mriedemgithub21:28
mnaserlol yeah21:28
mnasergithub21:28
mriedemnot cli21:28
dansmithheh nice21:28
mnasermriedem: i noticed if you refresh a few times21:28
mnaserit eventually goes through21:28
mnaseri assume that it caches portions of the git blame page21:28
mriedemgit blame nova.compute.manager21:28
mriedemor nova.virt.libvirt.driver21:28
mnaserand then eventually caches enough so the request doesnt timeout21:29
openstackgerritMerged openstack/nova master: Add "links" in the response of "nova show" for a down-cell instance  https://review.openstack.org/64030221:29
mriedemwell i see this has come up before https://review.openstack.org/#/q/I3b9f60a2c4aea904f39d6914c8f35c4d682a6b3a21:31
mriedemfollowed by https://review.openstack.org/#/q/I53d5284907d44ae8b5546993f8fd461b385c39e621:32
mriedemwe can't seem to make up our mind21:32
*** betherly has quit IRC21:36
*** irclogbot_2 has quit IRC21:38
*** betherly has joined #openstack-nova21:39
openstackgerritMatt Riedemann proposed openstack/nova master: WIP: Force refresh nw info cache if cache is empty but ports attached  https://review.openstack.org/64051621:40
*** tbachman has quit IRC21:40
mriedemmelwitt: well i'm kind of stuck on how to proceed with your counting quotas from placement stuff21:41
openstackgerritTim Rozet proposed openstack/nova master: Fixes race condition with privsep utime  https://review.openstack.org/62574121:44
mriedemstephenfin: in case you didn't notice it looks like mikal would like you to remove the -2 from https://review.openstack.org/#/c/554438/21:49
*** tbachman has joined #openstack-nova21:50
*** tbachman has quit IRC21:52
*** betherly has quit IRC21:52
*** betherly has joined #openstack-nova21:55
openstackgerritMerged openstack/nova master: conf: Deprecate 'disable_libvirt_livesnapshot' option  https://review.openstack.org/62693221:56
openstackgerritMerged openstack/nova master: Stop using "nova" in API samples when creating a server  https://review.openstack.org/63987421:57
openstackgerritMerged openstack/nova master: Make Claim._claim_test handle SchedulerLimits object  https://review.openstack.org/63641021:57
*** markvoelker has quit IRC22:01
*** wolverineav has quit IRC22:04
*** wolverineav has joined #openstack-nova22:05
*** betherly has quit IRC22:08
*** hongbin has joined #openstack-nova22:15
mriedemis InstanceNUMACell.id a db primary key or something from the actual host...?22:16
*** ivve has quit IRC22:16
mriedemi guess the latter22:17
*** awaugama has quit IRC22:22
melwittmriedem: reading through your comments22:32
*** jmlowe has joined #openstack-nova22:32
*** wolverineav has quit IRC22:35
*** awalende has quit IRC22:36
*** awalende has joined #openstack-nova22:36
melwittmriedem: why stuck? do you and dansmith think the quota counting should try to parse user_id out of request specs? I assumed not22:37
mriedemyou mean the InstanceMapping._from_db_object?22:37
mriedemstuck on whether or not InstanceMapping.user_id should be nullable, yes22:37
melwittoh22:39
melwittso the desire is to get it from the request spec json blob in _from_db_object22:39
*** tbachman has joined #openstack-nova22:39
mriedemwell we have 3 places to look22:40
mriedemrequest spec, build request, instance in the cell22:40
*** awalende has quit IRC22:40
mriedemif the cell is down, the last one is going to blow up22:40
melwittbuild request doesn't have it, AFAICT22:40
mriedemBuildRequest.instance.user_id22:40
mriedemBuildRequest.instance is a json blob22:40
melwittoh, ok22:40
melwittI didn't know that was persisted in the db22:41
mriedemyup22:41
mriedemhttps://github.com/openstack/nova/blob/master/nova/compute/api.py#L93422:42
mriedemelbragstad: i waxed policy rule names just for you https://review.openstack.org/#/c/621476/33/nova/policies/server_topology.py@2022:44
*** ajo_ has joined #openstack-nova22:45
*** odyssey4me_ has joined #openstack-nova22:45
*** jbryce_ has joined #openstack-nova22:45
*** betherly has joined #openstack-nova22:46
cfriesenany chance either one of you could take a quick look at https://review.openstack.org/#/c/620706/ ?   It's on a runway and nobody has looked at it since Matt and Alex on the 27th.22:46
mriedemcfriesen: i'll be honest i'm avoiding that one22:46
cfriesenheh22:46
cfriesengood to know22:46
mriedemit's just low priority for me22:46
mriedemcan the hardware stuff be split out of that?22:47
mriedemyou windriver folk seem to enjoy the larger single patches...22:47
mriedemthat would drop 1/3 of the size (and context to load) for the api part of it22:49
cfriesenyeah, I can break out the hardware bit.  you're thinking do that part first?22:49
mriedemit's a dependency for the api stuff it looks like22:49
cfriesenor second22:49
mriedemidk honestly22:49
cfriesenI think it could go either way, a bunch of the api stuff is just calling existing checks22:50
mriedemthe new hardware stuff is "Added the following validations to _validate_flavor_image():   cpu_policy and cpu_thread_policy values, serial ports, realtime   mask, cpu topology." right?22:50
*** wolverineav has joined #openstack-nova22:50
cfriesenno, much less than that22:51
mriedemoh i see https://review.openstack.org/#/c/620706/27/nova/compute/api.py@62822:51
cfriesenthose are all existing functions22:52
mriedemso leakypipes and bauzas approved that spec but haven't reviewed the code at all yet...22:52
*** eharney has quit IRC22:52
*** odyssey4me has quit IRC22:52
*** nicolasbock has quit IRC22:52
*** lamt has quit IRC22:52
*** jbryce has quit IRC22:52
*** melwitt has quit IRC22:52
*** ajo has quit IRC22:52
*** ajo_ is now known as ajo22:52
*** odyssey4me_ is now known as odyssey4me22:52
*** jbryce_ is now known as jbryce22:52
mriedemthis just seems like something i'd have to fine-toothed-comb through22:53
*** melwitt has joined #openstack-nova22:53
mriedemand i'm already at mental capacity for lots of other blueprints i'm reviewing right now22:53
mriedemplus it's fucking snowing again and i'm about to lose it22:53
mriedemhttps://www.kimt.com/content/news/February-has-been-a-record-breaking-month-for-snow-and-it-looks-like-March-will-start-with--506431011.html22:54
elbragstadmriedem mmm - interesting22:54
mriedemelbragstad: not really, but if that floats your boat :)22:55
mriedemi think it might22:55
*** lchabert has quit IRC22:55
elbragstadoh - you know it22:55
elbragstadwhat does host-info do?22:55
elbragstadlemme guess, gets information about the host?22:55
mriedemshows admin-sensitive details22:55
cfriesenmriedem: I get where you're coming from...but from my end it's frustrating since being on a runway is supposed to mean it gets reviews...22:56
mriedemthe api returns stuff that is admin-or-owner and if you're admin you get some extra juicy bits22:56
mriedemcfriesen: yup, i know22:56
elbragstadmmm22:56
elbragstadideally - it would be nice if the policies for those types of things went away22:56
mriedemelbragstad: what replaces them?22:57
*** lchabert has joined #openstack-nova22:57
mriedemdifferent route?22:57
elbragstadwell - i'm not sure if this will work based on the two minutes i've known about this functionality22:57
elbragstadbut22:57
elbragstadyou could populate those bits if the request was made with a system-scoped token + a certain role22:58
mriedemi knew you'd say something about system scoped tokens22:58
elbragstadmmmm22:59
*** eharney has joined #openstack-nova22:59
*** markvoelker has joined #openstack-nova22:59
mriedemdoes your wife tell you to stop talking about those at dinner?22:59
*** betherly has quit IRC22:59
elbragstadwe don't talk about policy at dinner22:59
mriedemjust baby stuff22:59
elbragstadi usually just fall asleep staring at a plate23:00
mriedemalright, anyway, not urgent, just mentioned it since i know it's a thing23:00
mriedemnot sure if the host_info is a subaction or attribute based on the policy naming doc23:00
mriedemseems like an 'attribute'23:00
elbragstadyeah - i suppose you could think of it that way, too23:00
elbragstadsinec it's a collection of attributes that only system users get to see, right?23:01
mriedemyeah this is definitely system scope stuff, not project admin23:01
mriedemb/c it's about the physical infra23:01
elbragstadsure23:01
openstackgerritya.wang proposed openstack/nova master: Select cpu model from a list of cpu models  https://review.openstack.org/63783423:01
elbragstadok - so i guess it depends on if you want to maintain two policies for the "same" API23:01
mriedemwe have some of that in other places23:02
elbragstador if you want to just implement a little filtering in the service and have a single policy to protect a single API23:02
mriedembased on scope type...23:02
elbragstadyes23:02
mriedemso rule is admin-or-owner, and you get the extras with a system scope token23:02
*** awalende has joined #openstack-nova23:02
elbragstadthe rule would be something like..23:03
elbragstad(role:admin and system_scope:all) or rule:owner23:03
mriedemok, which can still be overridden yeah?23:04
mriedemif you want non-system scope but support-level people to be able to see stuff23:04
elbragstadoperators can override - if that's the check string23:04
elbragstadsupport-level people?23:05
melwittmriedem: replied to your comment. I'll push another PS soon. thanks for raising the unresolved bit that I had missed23:05
elbragstadsupport-level people to me == someone with the reader role on the system (`openstack role add --user bob --system reader`)23:05
mriedempeople that aren't operators, they don't have the keys to the hardware, but they need some kind of admin-y level auth to see things23:05
elbragstadah23:05
mriedemelbragstad: yeah ok23:05
mriedemthat23:05
elbragstadright...23:05
elbragstadso you could rework that policy to be more flexible23:06
*** betherly has joined #openstack-nova23:06
elbragstadby default, to allow for those cases23:06
elbragstad(role:reader and system_scope:all) or rule:owner23:06
elbragstad^ should do it23:06
*** frickler has quit IRC23:06
mriedemok, well let's not go crazy here. i have to run pickup the kid before i get snowed in23:06
*** frickler has joined #openstack-nova23:06
elbragstadfwiw - we just started documenting a bunch of this23:06
elbragstadhttps://review.openstack.org/#/c/638563/23:06
elbragstad^ you know - in case you decide you want to read it over dinner23:07
mriedemi just might23:07
mriedemi talk shop at dinner and it just goes into the void23:07
elbragstadlol23:07
*** mriedem is now known as mriedem_afk23:08
*** betherly has quit IRC23:11
*** luksky has quit IRC23:15
*** betherly has joined #openstack-nova23:16
marstmriedem: Hey Matt! Just to confirm that fake nova driver works perfectly with minimal local.conf in stable/rocky devstack. Thanks a lot once again!23:19
*** betherly has quit IRC23:25
*** betherly has joined #openstack-nova23:29
*** markvoelker has quit IRC23:31
openstackgerritChris Friesen proposed openstack/nova master: Add emulated TPM support to Nova  https://review.openstack.org/63136323:31
*** awalende_ has joined #openstack-nova23:32
*** sapd1 has joined #openstack-nova23:33
*** awalende has quit IRC23:35
*** awalende_ has quit IRC23:36
*** awalende has joined #openstack-nova23:37
*** awalende has quit IRC23:41
*** betherly has quit IRC23:43
*** wolverineav has quit IRC23:55
*** wolverineav has joined #openstack-nova23:55

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!