Friday, 2021-03-26

« Thursday, 2021-03-25 Index Saturday, 2021-03-27 »
*** luksky has quit IRC00:05
*** mlavalle has quit IRC00:16
*** tosky has quit IRC00:26
*** LinPeiWen has joined #openstack-nova00:36
*** sapd1 has quit IRC00:48
*** hamalq has quit IRC01:04
*** k_mouza has joined #openstack-nova01:15
*** k_mouza has quit IRC01:20
*** bhagyashris has quit IRC01:39
*** jamesdenton has quit IRC01:49
*** jamesden_ has joined #openstack-nova01:49
*** rcernin has quit IRC02:46
*** rcernin has joined #openstack-nova02:52
*** macz_ has joined #openstack-nova03:05
openstackgerritMerged openstack/nova stable/stein: Use subqueryload() instead of joinedload() for (system_)metadata  https://review.opendev.org/c/openstack/nova/+/76181203:06
*** macz_ has quit IRC03:09
openstackgerritsean mooney proposed openstack/nova master: libvirt: Delegate OVS plug to os-vif  https://review.opendev.org/c/openstack/nova/+/60243203:12
*** k_mouza has joined #openstack-nova03:16
eanderssonmelwitt do you know what is going on at the rocky ci?03:16
eandersson> Cherry pick hash 9e9c022bde3a3ffdf0dd87e21bf9afde0dbc1e74 not on any master or stable branches03:16
sean-k-mooneythats form the cherry pick script we have to ensure we dont skip branches wehn we backport03:17
sean-k-mooneyits run as part of the pep8 job03:17
sean-k-mooneyeandersson: this has not merge in trian yet so all the previous version will fail03:18
sean-k-mooneythat has been cherry picked incorrectly03:18
sean-k-mooneyhttps://review.opendev.org/c/openstack/nova/+/356020 is merged03:19
sean-k-mooneyhttps://review.opendev.org/c/openstack/nova/+/782982 is pending03:19
sean-k-mooneybut we shoudl have the cherry picked form lines for each branch03:19
sean-k-mooneywell03:20
sean-k-mooneyits bit that its incorrect persay03:20
sean-k-mooneybut untill https://review.opendev.org/c/openstack/nova/+/782982 merges it cant pass pep803:20
*** k_mouza has quit IRC03:20
sean-k-mooneyeandersson: https://github.com/openstack/nova/blob/master/tools/check-cherry-picks.sh03:21
sean-k-mooneythat is what is causing it to fail03:21
*** hemanth_n has joined #openstack-nova03:22
sean-k-mooneyeandersson: im not sure why the first cherry pick start a train03:22
sean-k-mooneyi guess this merged a year ago so maybe train was correct03:23
sean-k-mooneyassumign it merged in ussuri03:23
sean-k-mooneyeandersson: after the train and stien version merge a recheck will allow it to pass03:25
sean-k-mooneyeandersson: so dont need to worry about it for now03:25
*** macz_ has joined #openstack-nova03:25
*** macz_ has quit IRC03:30
eanderssonSounds good03:37
eanderssonIt's just weird that it was under pep8 and the message for sure wasn't clear :D03:38
eanderssonWould never have figured that out.03:38
sean-k-mooneypep8 tradtionally was used for all linting jobs03:38
eanderssonAt least not without digging into the code.03:38
sean-k-mooneysome non python porject starte using a tox linters env03:38
sean-k-mooneybut nova and the rest of the old "core" project all had lintere in pep8 already so we just extend it03:39
eanderssonIs this the one you looked at btw? https://review.opendev.org/c/openstack/nova/+/76181303:39
eanderssonI re-checked it after the Stein backport was patched.03:40
eanderssonbut will try again just in case03:40
sean-k-mooneyno03:40
sean-k-mooneyi serched for 9e9c022bde3a3ffdf0dd87e21bf9afde0dbc1e7403:40
sean-k-mooneywhcih gave me https://review.opendev.org/q/9e9c022bde3a3ffdf0dd87e21bf9afde0dbc1e7403:40
eanderssonI knew it was a mistake to include the example id :D03:40
eanderssonI was considering replacing 9e9c with X or something03:41
eanderssonAnyway lets wait for the re-check and see I guess03:41
sean-k-mooneysame think there though 66a4ea9f25cdc75691ed55adc842b795c10b4191 was not merged when it ran03:42
eanderssonYea it's very likely I got trigger happy there.03:42
*** mkrai has joined #openstack-nova03:43
sean-k-mooneyits merge now though https://review.opendev.org/c/openstack/nova/+/761812 so the recheck should pass03:43
sean-k-mooneyhum hopfully the grenade job will pass with the recheck too03:45
*** macz_ has joined #openstack-nova03:46
sean-k-mooneylooks like a filure in the post test hook https://zuul.opendev.org/t/openstack/build/23129f86fb5146a084bb04f3eb9fcfec/log/logs/devstack-gate-post_test_hook.txt#9503:50
sean-k-mooneywhen testing ceph03:50
*** macz_ has quit IRC03:51
*** brinzhang has joined #openstack-nova03:52
*** tkajinam has quit IRC04:01
*** tkajinam has joined #openstack-nova04:01
*** macz_ has joined #openstack-nova04:07
*** macz_ has quit IRC04:11
*** sapd1 has joined #openstack-nova04:28
*** ratailor has joined #openstack-nova04:36
*** mkrai has quit IRC04:39
*** mkrai has joined #openstack-nova04:40
*** k_mouza has joined #openstack-nova04:46
*** k_mouza has quit IRC04:50
*** sapd1 has quit IRC05:01
openstackgerritMerged openstack/nova master: Fix check_instance_shared_storage() call  https://review.opendev.org/c/openstack/nova/+/78302605:03
*** mkrai has quit IRC05:23
*** mkrai_ has joined #openstack-nova05:23
*** sapd1 has joined #openstack-nova05:57
*** ircuser-1 has joined #openstack-nova06:00
*** sapd1 has quit IRC06:03
*** jangutter has joined #openstack-nova06:04
openstackgerritJosephine Seifert proposed openstack/nova stable/stein: Add config parameter 'live_migration_scheme' to live migration with tls guide  https://review.opendev.org/c/openstack/nova/+/78319906:04
*** jangutter_ has quit IRC06:07
*** Luzi has joined #openstack-nova06:11
gibidansmith: thanks for the night whatch and the rechecks. The fix is merged and the RC1 patch is updated now.06:11
gibibauzas: ^^06:11
*** zzzeek has joined #openstack-nova06:12
*** macz_ has joined #openstack-nova06:12
*** zzzeek has quit IRC06:16
*** macz_ has quit IRC06:17
*** zzzeek has joined #openstack-nova06:19
*** zzzeek has quit IRC06:26
*** zzzeek has joined #openstack-nova06:28
*** arne_wiebalck has joined #openstack-nova06:36
*** elod is now known as elod_afk06:45
*** luksky has joined #openstack-nova06:45
*** ftarasenko has joined #openstack-nova07:05
*** rpittau|afk has quit IRC07:05
*** vdrok has quit IRC07:05
*** masayukig has quit IRC07:05
*** masayukig has joined #openstack-nova07:06
*** rpittau|afk has joined #openstack-nova07:06
*** NobodyCam has quit IRC07:06
*** vdrok has joined #openstack-nova07:06
*** NobodyCam has joined #openstack-nova07:06
*** ricolin has joined #openstack-nova07:09
*** rcernin has quit IRC07:11
*** dklyle has quit IRC07:11
*** ricolin has quit IRC07:14
openstackgerritOpenStack Release Bot proposed openstack/nova stable/wallaby: Update .gitreview for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78319507:16
*** ricolin has joined #openstack-nova07:17
openstackgerritOpenStack Release Bot proposed openstack/nova stable/wallaby: Update TOX_CONSTRAINTS_FILE for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78321607:17
openstackgerritOpenStack Release Bot proposed openstack/nova master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78321707:17
openstackgerritOpenStack Release Bot proposed openstack/nova master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/nova/+/78321807:18
*** dpawlik6 has joined #openstack-nova07:22
*** zzzeek has quit IRC07:25
*** macz_ has joined #openstack-nova07:31
*** macz_ has quit IRC07:35
*** rcernin has joined #openstack-nova07:38
*** zzzeek has joined #openstack-nova07:40
*** rcernin has quit IRC07:42
*** zzzeek has quit IRC07:44
*** rcernin has joined #openstack-nova07:55
*** zzzeek has joined #openstack-nova07:58
*** rcernin has quit IRC07:59
*** xarlos has joined #openstack-nova08:03
*** zzzeek has quit IRC08:04
*** rpittau|afk is now known as rpittau08:09
*** rcernin has joined #openstack-nova08:13
*** rcernin has quit IRC08:17
openstackgerritBalazs Gibizer proposed openstack/nova stable/wallaby: [stable-only]Update .gitreview for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78319508:24
openstackgerritBalazs Gibizer proposed openstack/nova stable/wallaby: Update TOX_CONSTRAINTS_FILE for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78321608:26
openstackgerritBalazs Gibizer proposed openstack/nova stable/wallaby: [stable-only]Update TOX_CONSTRAINTS_FILE for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78321608:27
*** lpetrut has joined #openstack-nova08:28
*** bhagyash- has joined #openstack-nova08:31
gibisean-k-mooney: could you please hit this two automatic os-vif patches https://review.opendev.org/c/openstack/os-vif/+/781891 ?08:32
*** bhagyash- is now known as bhagyashris08:32
bauzasmorning08:38
bauzasgibi: sorry, looks like yesterday Frenode had some split08:38
bauzasand maybe you pinged me08:38
bauzasthanks for the RC1 patch08:39
*** zzzeek has joined #openstack-nova08:42
*** bauzas has quit IRC08:45
*** bauzas has joined #openstack-nova08:46
*** zzzeek has quit IRC08:47
*** elod_afk is now known as elod08:52
*** sapd1 has joined #openstack-nova08:53
gibibauzas: no worries everything went well during the night :)08:54
bauzasgibi: I'll need to be off this afternoon btw.08:54
gibiack08:55
gibinothing pressing is in the queue08:55
*** sapd1 has quit IRC08:57
*** swp20 has joined #openstack-nova08:58
openstackgerritMerged openstack/osc-placement master: Update master for stable/victoria  https://review.opendev.org/c/openstack/osc-placement/+/75132008:58
swp20gibi: morning, i find disk_filter is removed from U version, could u please give me some ref links about the reason?09:00
stephenfinswp20: It'll be in the release notes09:01
*** lucasagomes has joined #openstack-nova09:01
stephenfinswp20: https://docs.openstack.org/releasenotes/nova/train.html09:01
stephenfinThe CoreFilter, DiskFilter and RamFilter, which were deprecated in Stein (19.0.0), are now removed. VCPU, DISK_GB and MEMORY_MB filtering is performed natively using the Placement service.09:01
*** rcernin has joined #openstack-nova09:01
*** andrewbonney has joined #openstack-nova09:02
swp20stephenfin: thanks a lot.09:02
*** derekh has joined #openstack-nova09:04
*** rcernin has quit IRC09:06
openstackgerritMerged openstack/osc-placement master: Add Python3 wallaby unit tests  https://review.opendev.org/c/openstack/osc-placement/+/75132109:07
*** zzzeek has joined #openstack-nova09:10
swp20stephenfin: do we plan to move the remain filters to placement?09:12
*** zzzeek has quit IRC09:13
*** zzzeek has joined #openstack-nova09:15
*** dtantsur|afk is now known as dtantsur09:17
bauzasswp20: no, that depends09:19
bauzasswp20: some filters can be deprecated if we can create what we name 'prefilters' using the Placement API09:19
bauzasswp20: but some others, no09:19
swp20bauzas: got it, thanks.09:22
swp20bauzas: placemet has support PCI filter such as GPU/FPGA devices, isn't it?09:25
gibivgpu mdevs are modelled in placement but PCI devices not yet09:27
gibidue to qos min bandwidth scheduling support some sriov PFs are modelled in placement but that only used for the bandwdith resource at the moment09:27
gibithe PCI schduling still happens in the PCIFilter in nova09:28
*** kashyap has joined #openstack-nova09:30
*** kashyap has left #openstack-nova09:30
*** tosky has joined #openstack-nova09:30
*** kashyap has joined #openstack-nova09:30
*** grandchild has joined #openstack-nova09:30
bauzasswp20: we only model vGPUs09:30
bauzasswp20: but we don't really model PCI devices09:31
bauzasand given we also have NUMA PCI affinity, we would also need to support this09:31
bauzasa long story here.09:31
*** martinkennelly has joined #openstack-nova09:31
swp20gibi, buazas: got, thanks a lot.09:33
*** eharney has quit IRC09:34
openstackgerritOpenStack Release Bot proposed openstack/placement stable/wallaby: Update .gitreview for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323209:35
openstackgerritOpenStack Release Bot proposed openstack/placement stable/wallaby: Update TOX_CONSTRAINTS_FILE for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323309:35
openstackgerritOpenStack Release Bot proposed openstack/placement master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323409:35
openstackgerritOpenStack Release Bot proposed openstack/placement master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/placement/+/78323509:36
*** whoami-rajat has joined #openstack-nova09:44
*** hoonetorg has joined #openstack-nova09:49
*** eharney has joined #openstack-nova09:52
*** rcernin has joined #openstack-nova09:53
*** k_mouza has joined #openstack-nova09:59
*** rcernin has quit IRC10:00
*** rcernin has joined #openstack-nova10:05
*** rcernin has quit IRC10:10
*** macz_ has joined #openstack-nova10:10
*** sapd1 has joined #openstack-nova10:11
*** macz_ has quit IRC10:14
*** rcernin has joined #openstack-nova10:27
*** rcernin has quit IRC10:32
*** sapd1 has quit IRC10:35
*** k_mouza has quit IRC10:39
*** k_mouza has joined #openstack-nova10:39
*** rcernin has joined #openstack-nova10:42
*** k_mouza_ has joined #openstack-nova10:43
*** k_mouza has quit IRC10:44
*** ociuhandu has joined #openstack-nova10:47
*** rcernin has quit IRC10:52
*** macz_ has joined #openstack-nova10:52
*** macz_ has quit IRC10:56
*** rcernin has joined #openstack-nova11:00
*** ociuhandu has quit IRC11:02
*** rcernin has quit IRC11:05
*** gokhani has joined #openstack-nova11:13
*** macz_ has joined #openstack-nova11:13
*** sapd1 has joined #openstack-nova11:17
*** rcernin has joined #openstack-nova11:17
*** macz_ has quit IRC11:18
*** sapd1 has quit IRC11:22
*** rcernin has quit IRC11:22
kashyapgibi: or anyone else: Can I borrow your eyes: In this method here _add_rng_device() — do you see a way where 'hw_rng:allowed = False' on image property might *not* take effect? -- https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L5665,#L568211:25
openstackgerritSlawek Kaplonski proposed openstack/nova master: [neutron] Get only ID and name of the SGs from Neutron  https://review.opendev.org/c/openstack/nova/+/78327511:31
*** ociuhandu has joined #openstack-nova11:32
*** macz_ has joined #openstack-nova11:34
*** owalsh has quit IRC11:35
*** macz_ has quit IRC11:38
*** ociuhandu has quit IRC11:41
*** mkrai_ has quit IRC11:42
*** mkrai_ has joined #openstack-nova11:42
*** rcernin has joined #openstack-nova11:50
*** rcernin has quit IRC11:54
*** owalsh has joined #openstack-nova11:55
*** ociuhandu has joined #openstack-nova11:56
*** tbarron has joined #openstack-nova11:58
*** mkrai_ has quit IRC12:01
*** ociuhandu has quit IRC12:10
*** k_mouza_ has quit IRC12:13
*** ociuhandu has joined #openstack-nova12:22
gibikashyap: the image_meta is unused in that function12:24
gibionly the flavor extra_spec is used12:24
gibiso that function does not care about any image property12:24
kashyapgibi: Oh, so the "hw_rng:allowed" image meta is useless in this case then12:25
gibiseem so12:28
gibiit might be a bug12:28
kashyapHmm, OSP QE reported it to me...12:29
gibias the function gets an image_meta parameter but the it does not use it12:29
*** ociuhandu has quit IRC12:31
kashyapgibi: Right; I guess we should:12:31
kashyapflavor.extra_specs.get() or image_imeta.properties.get() ...12:31
gibiyeah that could be a solution12:32
sean-k-mooneywe have a function for that12:34
sean-k-mooneyat lesate wehre we named then the same12:34
sean-k-mooneyhw_rng:allowed12:35
sean-k-mooneyshoudl not be use in image meta12:35
sean-k-mooneyonly the flavor can allow it12:35
sean-k-mooneyhttps://github.com/openstack/glance/blob/master/etc/metadefs/compute-libvirt-image.json#L54-L59 is the only image metadata value12:37
kashyapsean-k-mooney: You mean "hw_rng:allowed" can only be used by extra_spec and not image_meta?12:38
sean-k-mooneyyes12:38
sean-k-mooneythat was a desgin chocie12:38
sean-k-mooneybecause we used to use /dev/random not /dev/urandom as you know12:39
sean-k-mooneythere were concerns about running out of entropy12:39
kashyapsean-k-mooney: Wait.  What you're pointing to is different: "hw_rng_model" --> this is useless now.12:39
sean-k-mooneyso it was an admin only extra spec12:39
sean-k-mooneywhich defaut to false12:39
kashyapsean-k-mooney: Okay ... so it's not a "bug" to not have it via image_meta12:39
sean-k-mooneyright it would be a bug to use it in image meta but the fact its not there is not12:40
sean-k-mooneyhw_rng_model is pretty useless not too yes12:40
sean-k-mooneyhttps://github.com/openstack/nova/blob/master/nova/objects/image_meta.py#L387-L39412:41
sean-k-mooneykashyap: but as you said we should not remove/deprecate it yet ^12:41
kashyapsean-k-mooney: You mean "as I said" in the commit message? :)12:41
sean-k-mooneyunless that comment you worte is also incorect12:41
kashyapsean-k-mooney: No; it's not incorrect.  I also wrote this in the commit message of the patch that introduced it:12:42
kashyap      - The 'hw_rng_model' image metadata property is now rendered12:42
kashyap        (temporarily) useless -- as it's not used anywhere outside the12:42
kashyap        _add_rng_device() method.  But we don't want to deprecate it yet, as12:42
kashyap        we may extend it (see code comment for details); docucment that.12:42
kashyapSo we wan to keep it.12:42
sean-k-mooneywell there too but i was not reading the comit message12:42
sean-k-mooneyi rember this form when the rng supprot was first beign added12:43
kashyapRigh; I wrote the code comment in image_meta.py12:43
*** gokhani has quit IRC12:43
*** hemanth_n has quit IRC12:43
*** ratailor has quit IRC12:44
*** ociuhandu has joined #openstack-nova12:45
*** gokhani has joined #openstack-nova12:49
*** ociuhandu has quit IRC12:50
*** ociuhandu has joined #openstack-nova12:52
*** ociuhandu has quit IRC12:57
*** rcernin has joined #openstack-nova12:57
*** rcernin has quit IRC12:58
*** rcernin has joined #openstack-nova12:58
openstackgerritBalazs Gibizer proposed openstack/osc-placement master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/osc-placement/+/78189412:58
*** ociuhandu has joined #openstack-nova13:00
*** k_mouza has joined #openstack-nova13:00
openstackgerritBalazs Gibizer proposed openstack/osc-placement master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/osc-placement/+/78189413:02
*** martinkennelly has quit IRC13:05
*** ociuhandu has quit IRC13:10
dansmithgibi: ah, based on  the time, it must have been the last one before I went to bed, cool13:14
gibidansmith: yepp you almost saw the patch landing13:19
gibi:)13:19
dansmithwell, sorry13:21
gibino worries. I'm glad you pushed the patch forwatrd13:22
*** macz_ has joined #openstack-nova13:39
openstackgerritMerged openstack/placement master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323413:40
*** ociuhandu has joined #openstack-nova13:40
*** macz_ has quit IRC13:43
*** k_mouza has quit IRC13:47
*** k_mouza_ has joined #openstack-nova13:47
*** rpittau is now known as rpittau|afk13:50
*** gokhani has quit IRC13:53
*** gokhani has joined #openstack-nova13:55
*** ociuhandu has quit IRC13:58
*** mlavalle has joined #openstack-nova13:58
openstackgerritBalazs Gibizer proposed openstack/nova master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/nova/+/78321814:04
gibisean-k-mooney: could you please hit this two automatic os-vif patches https://review.opendev.org/c/openstack/os-vif/+/781891 ?14:10
*** ociuhandu has joined #openstack-nova14:10
*** brinzhang has quit IRC14:14
openstackgerritMerged openstack/os-vif master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/os-vif/+/78189014:17
*** derekh has quit IRC14:20
*** ociuhandu has quit IRC14:21
openstackgerritMerged openstack/placement stable/wallaby: Update .gitreview for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323214:22
openstackgerritMerged openstack/placement stable/wallaby: Update TOX_CONSTRAINTS_FILE for stable/wallaby  https://review.opendev.org/c/openstack/placement/+/78323314:22
openstackgerritMerged openstack/placement master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/placement/+/78323514:23
openstackgerritBalazs Gibizer proposed openstack/nova master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/nova/+/78321814:23
*** smcginnis has quit IRC14:24
*** smcginnis has joined #openstack-nova14:26
*** dklyle has joined #openstack-nova14:29
*** macz_ has joined #openstack-nova14:31
*** ociuhandu has joined #openstack-nova14:33
sean-k-mooneygibi: ill take a look now14:34
sean-k-mooneygibi: a gmann  has don it alredy14:34
gibisean-k-mooney: thanks14:34
gibiohh14:34
sean-k-mooneygmann++14:35
gibithanks then14:35
*** macz_ has quit IRC14:36
*** ociuhandu has quit IRC14:38
*** jangutter_ has joined #openstack-nova14:39
*** ociuhandu has joined #openstack-nova14:42
*** jangutter has quit IRC14:42
*** ociuhandu has quit IRC14:44
*** ociuhandu has joined #openstack-nova14:44
*** macz_ has joined #openstack-nova14:48
*** derekh has joined #openstack-nova14:59
*** ociuhandu has quit IRC15:04
*** Luzi has quit IRC15:05
*** rcernin has quit IRC15:06
artomsean-k-mooney, hey, in https://review.opendev.org/c/openstack/nova/+/783084/1/nova/tests/functional/libvirt/test_pci_sriov_servers.py#382 when you say it's a side effect of our Neutron fixture, what did you mean?15:08
artomBtw, that should have been WIP :P15:08
artomI'm just starting to explore wth's going on15:08
sean-k-mooneyi mean that in reality you will not get that key error15:08
sean-k-mooneythe pci_slot key will be set in the profile15:09
sean-k-mooneyit just will have the wrong value15:09
sean-k-mooneyso your test is failing but not for the correct reason15:09
sean-k-mooneydid you see my top level comment15:09
artomsean-k-mooney, right, so... 1. what bit in the fixture isn't like reality? and 2. dvd seems to have gotten the same keyerror here: https://bugzilla.redhat.com/show_bug.cgi?id=1767797#c1715:09
openstackbugzilla.redhat.com bug 1767797 in openstack-nova "When unshelving an SR-IOV instance, the binding profile isn't reclaimed or rescheduled, and this might cause PCI-PT conflicts" [High,Assigned] - Assigned to alifshit15:09
sean-k-mooneyi suggested how to write the test diffeerntly15:09
artomsean-k-mooney, yeah, I actually started like that, and then went back because... well, I wasn't thinking straight. Lemme do it "properly", and upload that I find15:10
sean-k-mooneyi think that was self inflicted15:11
sean-k-mooneyyou can actully unshleve i hvae tested that15:11
sean-k-mooneyso the nasty part of this bug is it appears to work but its subtely broken15:11
valleedelislesean-k-mooney, can we unshelve on a host with the same pci device in use?15:11
artomvalleedelisle, oh that's your upstream nick15:12
valleedelislevalleedelisle, I think that the point of that bz is the pci device not being re-calculated15:12
valleedelisleyeah dvd is already in use hehe15:12
sean-k-mooneyvalleedelisle: no that will cause the libvirt error in the upstream bug15:12
sean-k-mooneylibvirt/qemu will cay the device is already in use15:12
sean-k-mooneybut if the pci device is not in use then yes15:12
artomvalleedelisle, 'dvd :No such nick/channel'15:12
valleedelislesean-k-mooney, yeah but shouldn't we recalculate pci device ?15:12
valleedelisleartom, it's registered15:12
sean-k-mooneywe do15:12
artomBleah :(15:13
sean-k-mooneyand we claim it in the db15:13
sean-k-mooneywe just dont update the port profile15:13
sean-k-mooneywhich is the bug15:13
valleedelisleack15:13
sean-k-mooneyi repodcued the bug manually on my sriov hardware at home and confirmed that i just did not have time to work on a fix a the time15:13
openstackgerritMerged openstack/nova master: Update master for stable/wallaby  https://review.opendev.org/c/openstack/nova/+/78321715:15
*** sapd1 has joined #openstack-nova15:16
*** lpetrut has quit IRC15:19
*** Hazelesque has quit IRC15:31
*** Hazelesque has joined #openstack-nova15:34
artomsean-k-mooney, btw, in order to actually write the unshelve func test properly, I need to fix the Neutron fixture to go further than that KeyError15:39
*** ociuhandu has joined #openstack-nova15:43
gibisean-k-mooney: you will probably like this https://storyboard.openstack.org/#!/story/200876415:46
sean-k-mooneygibi: well its on story borad but lets see15:46
sean-k-mooneyoh ya i think it has uses15:46
sean-k-mooneyya preventing loops and orpans is something we woudl have to do15:47
sean-k-mooneyto mantain the DAG stucture15:47
gibisean-k-mooney: how can we make orphans?15:47
gibisean-k-mooney: loop check is already implemented15:47
sean-k-mooneysettign parent to None15:47
gibithat does not orphan things that creates a new root provider15:48
sean-k-mooneyor a rp that does not exist15:48
sean-k-mooneywell it would be a root provider yes15:48
gibiyeah, non existing RP needs to be rejected (an rejected already)15:48
sean-k-mooneybut if you did that to your bandwith RPs it woudl be kind of broken15:48
gibibroken from nova an neutron perspective but not from placement perspective15:48
sean-k-mooneyyep i just commenting on jays comment you copied15:48
sean-k-mooneywell the TODO15:49
gibiplacement does not know the intention of the clients15:49
sean-k-mooneygibi: yep it would be a logic error15:49
sean-k-mooneybut not invalid in general15:49
gibiyepp from placement perspective creating a new root RP is valid15:49
sean-k-mooneyso used correctly i think its a useful feature15:49
gibiyepp15:49
sean-k-mooneyi also think its slightly diffrent then reshaping15:50
gibiit is15:50
gibireshape is to move allocations around15:50
gibithis is to move RPs around :)15:50
sean-k-mooneyallocation and inventores and potentialy creating more or less RPs15:50
*** jamesden_ has quit IRC15:50
gibireshape does not create RPs15:50
sean-k-mooneywhere as moving an RP is just that changing it postion in the grap15:50
gibibut yes it move inventories and allocations15:50
sean-k-mooneygibi: well it does not on the placment side15:51
sean-k-mooneybut our nova reshapes do15:51
gibinova creates RPs, and then reshape to those RPs15:51
sean-k-mooneyfor example the vgpu reshpe15:51
sean-k-mooneyyep15:51
gibi:)15:51
sean-k-mooneywe do multipel operation in the nova fucnitons that we incorreectly refer to as reshaping as a whole15:51
*** jamesdenton has joined #openstack-nova15:51
sean-k-mooneybut ya reshape in the placment api is a specific thing15:52
gibiyepp15:52
sean-k-mooneymy original use case for this was moving pci device if we later added numa nodes15:52
sean-k-mooneyor vgpus15:52
gibiyepp, I added that to the story as a reason15:53
sean-k-mooneye.g. just changing the partent form the root rp to the numa ndoe15:53
gibiexactly15:53
gibithat will be a reshape in nova but a reparent in placement :)15:53
sean-k-mooneythe only thing that woudl be invalidated by that would be the allocation summaries15:53
sean-k-mooneyso as long as we are not boot a vm or moveign one15:53
sean-k-mooneyit hink it woudl be ok15:53
artomstephenfin, that infamous comment of your for the cold migration functional test mocking is getting around15:53
artom[artom@zoe nova]$ ag dumb nova/tests/functional/libvirt/ | wc -l15:54
artom1115:54
stephenfinthe dog15:54
sean-k-mooneyartom: which one?15:54
sean-k-mooneyim surprised we have less use fo dumb in our func test :)15:54
artom# TODO(stephenfin): The mock of 'migrate_disk_and_power_off' should probably be less...dumb15:55
sean-k-mooneythere are many silly things15:55
sean-k-mooneyhehe15:55
stephenfinI did try to fix that one time15:55
stephenfinand quickly gave up15:55
sean-k-mooneyi dont think i have actully seen that beofre or at least read it15:55
stephenfintoo much involved15:55
sean-k-mooneybut im sure it could be15:55
*** LinPeiWen has quit IRC16:00
openstackgerritMerged openstack/os-vif master: Add Python3 xena unit tests  https://review.opendev.org/c/openstack/os-vif/+/78189116:03
*** ociuhandu has quit IRC16:08
openstackgerritArtom Lifshitz proposed openstack/nova master: WIP: Test SRIOV port move operations with PCI conflicts  https://review.opendev.org/c/openstack/nova/+/78308416:20
*** ociuhandu has joined #openstack-nova16:39
*** gyee has joined #openstack-nova16:40
*** hamalq has joined #openstack-nova16:43
*** hamalq_ has joined #openstack-nova16:44
*** ociuhandu_ has joined #openstack-nova16:44
*** ociuhandu_ has quit IRC16:44
openstackgerritArtom Lifshitz proposed openstack/nova master: WIP: Test SRIOV port move operations with PCI conflicts  https://review.opendev.org/c/openstack/nova/+/78308416:45
openstackgerritArtom Lifshitz proposed openstack/nova master: WIP: Neutron fixture: don't clobber profile and vif_details if empty  https://review.opendev.org/c/openstack/nova/+/78338516:45
artomsean-k-mooney, ^^ there :) Good time to pause for lunch16:45
*** hamalq has quit IRC16:47
*** ociuhandu has quit IRC16:48
*** lucasagomes has quit IRC16:58
*** artom has quit IRC17:02
*** jangutter_ has quit IRC17:08
*** jangutter has joined #openstack-nova17:08
*** iurygregory has quit IRC17:12
*** iurygregory has joined #openstack-nova17:13
*** artom has joined #openstack-nova17:18
sean-k-mooneyartom: yep that is a good approch17:25
sean-k-mooneythe inner move_operation fucntion true me for a sec but it makes sense17:25
artomsean-k-mooney, yeah, it's not great, but I wanted to test cold migration too17:26
sean-k-mooneythis is where i woudl like to use  named lambda but those are not allowed17:26
artomThat one works, thankfully17:26
sean-k-mooneyas far as i am aware this cannot happen for live mifgation since we dont use the claims in the same way17:26
sean-k-mooneyso ya17:27
sean-k-mooneyjust cold migratoien and shelve17:27
sean-k-mooneyalthough evacualte might also be affected17:27
artomIn fact for unshelve we don't have any claims17:27
artomI guess when it was written it was supposed to never change hosts?17:28
sean-k-mooneywell we do17:28
sean-k-mooneywe have pci claims in the db17:28
artomI dunno, but the way I read it, we just plop the instance on the compute, no questions asked17:28
sean-k-mooneyin the pci tracker17:28
sean-k-mooneywe do not use move claims17:28
*** dtantsur is now known as dtantsur|afk17:28
sean-k-mooneybecause its not actully a move operations17:28
sean-k-mooneyunshelve basically uses spawn17:29
sean-k-mooneythe vm never really exist in two compute hosts at the same time liek with migrrate and live migrate or evacuate17:29
sean-k-mooneyartom: we use the normal instance claim wich is created on the compute node17:30
openstackgerritStephen Finucane proposed openstack/nova master: conf: Deprecate AggregateImagePropertiesIsolation opts  https://review.opendev.org/c/openstack/nova/+/78339617:31
stephenfinsean-k-mooney: ^17:31
sean-k-mooneycool too bad that didnt happen 2 weeks ago17:31
sean-k-mooneyill take a look shortly17:31
sean-k-mooneyartom: i dont know if you have looked at unshelve but we go to task state spawning https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L652517:32
sean-k-mooneythen we do a normal instance claim17:33
sean-k-mooneyand call driver.spwan17:33
sean-k-mooneyhttps://github.com/openstack/nova/blob/master/nova/compute/manager.py#L6577-L658517:33
sean-k-mooneyartom: so we use https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/compute/claims.py#L4917:33
sean-k-mooneyfor unshelve just like we do for a new boot17:33
sean-k-mooneyartom: by the way im not sure why you need https://review.opendev.org/c/openstack/nova/+/783385/1/nova/tests/fixtures.py17:36
sean-k-mooneyyou are removing the copies that we make which im not sure is correct17:36
sean-k-mooneyother wise what you wote is logically the same17:37
artomsean-k-mooney, that fixes the KeyError17:38
sean-k-mooneyi mean if it works cool i just dont know why it does17:38
sean-k-mooneyright but body['port'].get('binding:profile') or {}17:38
sean-k-mooneyshoudl do the same thing17:38
sean-k-mooneyas17:38
sean-k-mooney            if body['port'].get('binding:profile'):17:38
sean-k-mooney                update['profile'] = body['port']['binding:profile']17:38
sean-k-mooneywell not quite it wont set {}17:39
sean-k-mooneyif there is no profile17:39
sean-k-mooneybut that hsoudl not clober things17:39
artomsean-k-mooney, right, but it means that if we update with no binding:profile, we'll hit the 'or {}', and assign that17:39
sean-k-mooneyoh i see whats hapening17:40
artomWhich effectively clobbers17:40
sean-k-mooneyya17:40
sean-k-mooneythat is not how the real api work17:40
sean-k-mooneywhat the real api does is keep the value of the fiedl you dont touch17:40
sean-k-mooneywhcih your change will do17:40
sean-k-mooneythe or {} shoudl have been more like17:41
*** yoctozepto has quit IRC17:41
sean-k-mooneyor self._port_bindings[port_id][host]['profile]17:41
sean-k-mooneyya ok this makes sense to me17:41
sean-k-mooneyartom: looks like nova.tests.functional.test_nova_manage.TestNovaManagePlacementHealPortAllocations.test_heal_multiple_port_allocations_rollback_fails17:46
sean-k-mooneybroke but oterwise this looks good17:46
sean-k-mooneystephenfin: nic commit message :)17:47
sean-k-mooneygood point on the hw_ possiblity but ya they will still work17:48
*** yoctozepto has joined #openstack-nova17:51
*** slaweq has quit IRC17:58
*** k_mouza_ has quit IRC17:58
*** slaweq has joined #openstack-nova18:03
*** slaweq has quit IRC18:09
*** jamesdenton has quit IRC18:09
*** jamesdenton has joined #openstack-nova18:10
stephenfinsean-k-mooney: That's a good point RE: using the isolated aggregates filtering feature via placement instead18:18
stephenfinI wonder if we should just just deprecate both of these filters instead?18:19
sean-k-mooneyi would be ok with that18:23
sean-k-mooneyptg topic maybe18:23
stephenfingood idea18:23
* stephenfin adds it18:23
openstackgerritStephen Finucane proposed openstack/nova master: conf: Deprecate AggregateImagePropertiesIsolation opts  https://review.opendev.org/c/openstack/nova/+/78339618:24
sean-k-mooneythe draw back is you might need to update your flavors/image with teh traits18:24
* artom was playing Go Fish with his daughter18:24
stephenfinYes, I'm not sure we could ever remove it18:24
sean-k-mooneybut at least of new deployment i would not recommend turnning them on any more na just use that18:24
artomRemote school is... interesting. She's already finished everything for the week with mom18:25
sean-k-mooneywell we could but it would need a few cycles18:25
stephenfinat least not without offering a mechanism to update stored image/flavor metadata for an existing instance18:25
stephenfinbut the code costs very little to maintain and could happily be ignored18:25
sean-k-mooneyartom: enjoy18:25
* artom is digging into the instance_claim on unshelve -> spawn()18:25
stephenfinanyway, PTG topic18:25
sean-k-mooneyartom: its the same type of claime we do for normal spawn18:26
artomSo yeah, we do a claim, but _update_port_binding_for_instance() assumed that pci_mapping can only change if there's a migration18:26
sean-k-mooneywhich include claiming the pci devices18:26
artompci_mapping = self._get_pci_mapping_for_migration(instance, migration)18:26
sean-k-mooneyartom: ya that should like what might be happening18:26
artomYeah, we claim them18:26
sean-k-mooneyah cool you found the issue then18:26
artomBut way before that, we've already updated the port bindings18:26
artomWith the old pci_slot, since the instance isn't actually migrating18:27
artomOr rather, we haven't actually done the update, since pci_slot doens't change18:27
* artom tries to think of a way to fix that that's backportable...18:27
sean-k-mooneywell we just need to update the port again18:28
sean-k-mooneyit would be nice not to do it twice18:28
sean-k-mooneybut that is the simiple backportable fix18:28
sean-k-mooneyas long as we update it before we generate the xml its fine18:29
sean-k-mooneythen eventually i would like to use the adress form the claim/db instead but that harder18:29
* artom wonder how we do it for normal boot18:30
artomIOW, how do we *set* the pci_slot in the first place18:31
artom_update_port_binding_for_instance() will only update it from a migration18:31
*** sapd1 has quit IRC18:33
artomAha _populate_neutron_binding_profile()18:38
*** mloza has joined #openstack-nova18:38
sean-k-mooneyya it happens in the port bidning right18:38
sean-k-mooneyartom: i think its this https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L1332-L133318:38
sean-k-mooneywe call that during live migration fore example https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/conductor/tasks/live_migrate.py#L41418:40
sean-k-mooneyartom: if you called self.network_api.bind_ports_to_host here https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L657918:43
sean-k-mooneyright before spawn i think that woudl fix it18:43
sean-k-mooneyyou would need to change18:44
sean-k-mooney with self.rt.instance_claim(context, instance, node, allocations,18:44
sean-k-mooney                                        limits):18:44
sean-k-mooneyto  with self.rt.instance_claim(context, instance, node, allocations,18:44
sean-k-mooney                                        limits)18:44
sean-k-mooneyas claim18:44
sean-k-mooneyand then you could use the claim to get the pci diceis i think18:45
*** andrewbonney has quit IRC18:47
sean-k-mooneyartom: the update would look like this https://opendev.org/openstack/nova/src/branch/master/nova/compute/manager.py#L10634-L1065718:49
sean-k-mooneyhum actullyno18:49
sean-k-mooneywell no this is how we actully do it for live miggration https://opendev.org/openstack/nova/src/branch/master/nova/compute/manager.py#L10597-L1063218:51
* artom wonders if we could just allocate_for_instance instead of _update_port_bindings_for_instance()...19:01
artomI guess we'd have to distinguish between unshelve from just SHELVED and also SHELVED_OFFLOADED19:01
sean-k-mooneyartom: in the normal spawn workflow after we take the instance calim we call  _build_resources19:01
sean-k-mooneywhich does call that yes19:01
sean-k-mooneywell it all alocate_netork19:02
sean-k-mooneyartom: https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/compute/manager.py#L234919:02
sean-k-mooneythat is what allocate the networks on spawn normally19:02
artomYeah, I've been looking at that...19:03
sean-k-mooneybut we dont do that on unshelve19:03
artomWhich makes sense... but only if the instance was only shelved19:03
sean-k-mooneywell we woudl be calling it form unshelve19:03
*** rcernin has joined #openstack-nova19:03
sean-k-mooneyhere https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L657919:03
sean-k-mooneylitrally after the instance_claim but just before spawn19:04
*** ralonsoh has quit IRC19:04
sean-k-mooneythis is where its update for other move ops right https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L3394-L341919:06
sean-k-mooneybut since thre is no migration that oviously does not work19:07
*** rcernin has quit IRC19:07
artomAh, we can't allocate_for_instance on unshelve because even OFFLOADED instances keep their ports19:09
artomSo I'd think all kinds of things would explode19:09
artomWe really do need just the _update, but we need to somehow pull the PCI info from the claim, and pass that to the neutron api later19:09
sean-k-mooneymaybe19:09
artom*layer19:09
sean-k-mooneyits doing more then we want anyway19:09
sean-k-mooney _update_ports_for_instance is closer19:10
sean-k-mooneyhttps://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L1498-L152419:11
sean-k-mooneythis is what actully update the port with the bindingprofile19:12
sean-k-mooneythe issue is we dont have a way to map the neutorn por tto the pci_request_id19:12
sean-k-mooneyif we had that we woudl jsut get the dev doing19:12
sean-k-mooney pci_devices = pci_manager.get_instance_pci_devs(19:12
sean-k-mooney                instance, pci_request_id)19:12
sean-k-mooneythen https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L1487-L149619:13
artomsean-k-mooney, yeah, I'm hitting that exact issue19:13
artomWith the migration, we have the mapping from old to new19:13
sean-k-mooneywould get that for us19:13
sean-k-mooneyyep19:13
sean-k-mooneyso i looked at sotring the pci resuet id in the vif breilfy19:13
sean-k-mooneybut unfortunetly that is a object change19:14
sean-k-mooneyhum https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L211219:14
artomI'm assuming pci_vendor_info and physical_network from the port binding isn't enough19:14
artomEven assuming we have the latter19:15
sean-k-mooneyhttps://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L216519:15
sean-k-mooneythe pci request id shoudl be the nueturno port uuid19:15
artomOhhh19:15
sean-k-mooneyso i thnk we can look it up like that?19:16
sean-k-mooneyso if we get teh port  we could loop over therm calling https://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L1498-L152419:18
sean-k-mooneypassing the port['id'] as pci_request_id19:18
sean-k-mooneyand then after that has updated them in memeory we woudl do a port update?19:18
sean-k-mooneythen i think we cloud call driver .spawn19:19
sean-k-mooneyartom: do you think that would work?19:20
sean-k-mooney... its not the neutron port uuid19:31
artomsean-k-mooney, think so, yeah19:31
artomLemme try it19:31
sean-k-mooneyartom: the requester_id is the neutron port uuid19:33
sean-k-mooneythe pci_request_id is the request_id19:33
sean-k-mooneyunfortunetly the requester_id is not sotred in teh db19:33
sean-k-mooneythe requested_networks network object i think is the only thing that has the mapping19:34
sean-k-mooneyi think this is the same roadblock i hit the last time19:35
sean-k-mooney...19:39
sean-k-mooneyso the instance_extra tbale has it but not fully19:39
sean-k-mooney[{"count": 1, "spec": [{"physical_network": "public"}], "alias_name": null, "is_new": false, "numa_policy": "preferred", "request_id": "03e6f4be-dc49-4825-a33d-337f2708be22", "requester_id": null}19:39
sean-k-mooneythe requester_id shoudl be the neutorn port uuid19:40
*** tkajinam_ has joined #openstack-nova19:40
sean-k-mooneyif it was we coudl get the request id and then look up the claimed deivce in the pci device table and mapt them back to the ports19:40
*** tkajinam has quit IRC19:41
sean-k-mooneyartom: i think im going to call it a day there. if we fixed that so the requester id was always set then we could use that19:43
sean-k-mooneyi think it curently not set becasue we dont save this back to the db and only pass it around19:43
sean-k-mooneyhttps://github.com/openstack/nova/blob/68ad2c67d891935c66e8c0e5a368695c202a9f49/nova/network/neutron.py#L2154-L216419:43
*** slaweq has joined #openstack-nova19:43
artomsean-k-mooney, ack yeah, my brain is getting rusty as well19:44
sean-k-mooneywe might need to do this in two parts19:47
sean-k-mooneyone patch that fixes populating that value and tries to file it in for exsitng instnace19:47
sean-k-mooneymaybe using the pci_slot19:47
sean-k-mooneyto figure out thte mappign between port and exctra19:48
sean-k-mooneyand tehn a secdond to use it to update the port or something like that19:48
sean-k-mooneyanyway enjoy your weekend19:48
sean-k-mooneywhat frustrating abtout this is every time i look at this we have almsot all the info we need19:49
sean-k-mooneyalmost19:49
*** gokhani has quit IRC19:49
*** k_mouza has joined #openstack-nova20:33
*** k_mouza has quit IRC20:37
*** amodi has quit IRC20:49
*** tbachman_ has joined #openstack-nova21:05
*** tbachman has quit IRC21:06
*** tbachman_ is now known as tbachman21:06
*** jangutter has quit IRC21:22
*** jangutter has joined #openstack-nova21:23
*** whoami-rajat has quit IRC21:30
*** derekh has quit IRC21:30
*** jamesdenton has quit IRC21:36
*** jamesden_ has joined #openstack-nova21:36
*** luksky has quit IRC23:14
*** gyee has quit IRC23:31
*** macz_ has quit IRC23:32
*** gyee has joined #openstack-nova23:32
*** tosky has quit IRC23:39
*** macz_ has joined #openstack-nova23:48
*** macz_ has quit IRC23:53
« Thursday, 2021-03-25 Index Saturday, 2021-03-27 »

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!