Thursday, 2022-04-21

opendevreviewMerged openstack/nova stable/victoria: [stable-only] Drop lower-constraints job  https://review.opendev.org/c/openstack/nova/+/83803204:05
*** amoralej|off is now known as amoralej06:47
kashyapgibi: Hi, will look at the review comments; if isomething is unclear, I'll ask 07:44
kashyapAnd thanks for the review!07:50
kashyapAlso, bonus points for reading the full commit message on baselineHypervisorCPU() :)07:51
gibi:)08:02
gibiUggla: added some extra information about the notification work in https://review.opendev.org/c/openstack/nova-specs/+/833669 let me know if you have questions08:26
Ugglagibi, thx I will have a look.08:28
opendevreviewribaudr proposed openstack/nova master: [WIP] Attach Manila shares via virtiofs (db+object)  https://review.opendev.org/c/openstack/nova/+/83119308:42
opendevreviewribaudr proposed openstack/nova master: [WIP] Attach Manila shares via virtiofs (manila abstraction)  https://review.opendev.org/c/openstack/nova/+/83119408:42
opendevreviewribaudr proposed openstack/nova master: [WIP] Attach Manila shares via virtiofs (drivers)  https://review.opendev.org/c/openstack/nova/+/83309008:42
opendevreviewribaudr proposed openstack/nova master: [WIP] Attach Manila shares via virtiofs (api)  https://review.opendev.org/c/openstack/nova/+/83683008:42
opendevreviewBalazs Gibizer proposed openstack/nova master: Record SRIOV PF MAC in the binding profile  https://review.opendev.org/c/openstack/nova/+/82924808:52
opendevreviewMerged openstack/nova master: Follow up for nova-manage image property commands  https://review.opendev.org/c/openstack/nova/+/83089509:43
opendevreviewBalazs Gibizer proposed openstack/nova stable/wallaby: Reproduce bug 1945310  https://review.opendev.org/c/openstack/nova/+/81141409:45
opendevreviewBalazs Gibizer proposed openstack/nova stable/wallaby: Query ports with admin client to get resource_request  https://review.opendev.org/c/openstack/nova/+/81141609:45
sean-k-mooneybauzas: if i have missed anythin or missrepresneted can you comment on http://lists.openstack.org/pipermail/openstack-discuss/2022-April/028268.html10:22
sean-k-mooneyim 99% sure that everything i said there is correct but i have not personally configured vGPUs so if you can confirm that woudl be good10:23
bauzassean-k-mooney: I already replied ;)10:58
bauzassean-k-mooney: but thanks for your reply 10:58
kashyapgibi[m]: Got a minute here?  I'm going a bit mad with the unit test -- something is wrong here because the 'obj' is printing empty <cpu/> when I do a print (obj) - https://paste.opendev.org/show/b9TodzDYO7ukIBO8mGEs/11:32
opendevreviewMerged openstack/nova master: Sync rootwrap.conf from oslo.rootwrap  https://review.opendev.org/c/openstack/nova/+/82322911:34
gibikashyap: will look in a minute11:35
kashyapNo rush11:37
*** bhagyashris_ is now known as bhagyashris11:55
opendevreviewWenping Song proposed openstack/nova-specs master: Usage of new trait of OWNER_NOVA  https://review.opendev.org/c/openstack/nova-specs/+/81951012:09
gibikashyap: now looking :)12:23
kashyapCool; I'm sure it's something stupid that I'm not parsing myself :D12:24
sean-k-mooneygibi: just pushed my comment for https://review.opendev.org/c/openstack/nova-specs/+/819510 but i see the also just pushed an update...12:27
gibikashyap: at least I can reproduce your problem locally :)12:29
gibisean-k-mooney: I will look back to the spec12:30
kashyapgibi: Yeah.  I wonder where I'm messing up the parsing here12:30
sean-k-mooneysongwenping_: can you look at my comment on patchset 512:31
sean-k-mooneysongwenping_: if you can respin the spec with those changes i think it would me mergable, please also include my version of the upgrade impact section12:32
sean-k-mooneyits imporant to call out the depenecy on upgrading placement before nova so that the new standard traits are aviable12:33
gibikashyap: your test code calls https://github.com/openstack/nova/blob/dc30a7e6d19509ce34cb89ffcdb4ff2c36469d89/nova/virt/libvirt/config.py#L75 that is pretty empty12:33
gibiwait, no12:34
gibihang on12:34
* kashyap hangs on12:34
gibi:D12:35
gibiit correctly calls nova.virt.libvirt.config.LibvirtConfigCPU.parse_dom12:35
kashyapYea, did you put a breakpoint() to see it12:37
*** amoralej is now known as amoralej|lunch12:38
kashyapAll I want to assert is that the test is parsing the <cpu> element correctly.  So assert for the presence of 'model' / 'vendor' etc?12:39
gibidoes nova use LibvirtConfigCPU to parse xml? I only see existing tests that verify to_xml but no parse_dom12:42
gibiit seems that parsing code in LibvirtConfigCPU.parse_dom is either incorrect or the example XML you use is incorrect12:45
kashyapgibi: Yeah, I didn't find any of parse_dom either12:45
gibithe parser code assumes that everything is directly under <cpu> but in your example XML the tags are under cpu/mode12:46
kashyapgibi: The example XML is ... trimmed down output from running `virsh domcapabilities` on my computer12:46
* kashyap goes to recheck12:46
gibiso the exta nesting by "mode" is makes the parse_dom to find nothing to match12:46
kashyapgibi: Run `virsh domcapabilities` and you'll see tags under several modes12:47
* kashyap checks12:47
gibithen you need to handle the extra mode nesting in the parse_dom code12:47
gibibut be aware that the to_xml code also emits the xml without the extra mode nesting right now12:48
kashyapgibi: Hmm, I see.  So, here's a test you do: the "everything under <cpu>" from `virsh capabilities`12:48
kashyapSo carefully run these two commands and compare the <cpu> bits:12:49
kashyap- virsh capabilities # notice the <cpu> — everything is under it; and it has no "policy" for features12:49
kashyap- virsh domcapabilities # notice the <cpu> and compare w/ the above12:49
gibiyeah I see12:50
gibiso capabilities emits different structure than domcapabilities12:50
kashyapYes, indeed.12:50
gibiso probably we need a different config object to parse the two different model12:50
gibias I assume nova still uses getCapabilities even after we start using getDomCapabitilies12:51
gibior can we simply ignore the mode part of the domcapabilities?12:51
gibias we can assume that mode is host-model12:51
* kashyap taps on the table and thinks12:52
gibiI have to step away from the keyboard for a while now but I will be back later this afternoon12:52
kashyapSure, no prob12:52
kashyapI think we can ignore the mode part here, as we indeed know that it's of 'host-model'12:55
kashyapAnd that works12:55
kashyapgibi: When you're back: indeed, putting the stuff under <cpu> does resolve the test prob.  (Also, near as I see, I don't think we need a diff config object)12:57
sean-k-mooneyform a consitncy point of view we should have different config objects13:04
sean-k-mooneywe have one config object per element type in the xml13:04
sean-k-mooneyso we shoudl not use the same python object to parse both the dom caps and normal caps13:05
sean-k-mooneykashyap: ^13:05
sean-k-mooneyso i agree with gibi here13:05
kashyapsean-k-mooney: There already exists a config object for domCaps13:06
kashyapAlso, Gibi was wondering out loud, and was not asking.  I'll dig a bit more and see what's required here.13:06
sean-k-mooneyack 13:06
opendevreviewMerged openstack/nova stable/xena: Reattach mdevs to guest on resume  https://review.opendev.org/c/openstack/nova/+/82112613:13
*** amoralej|lunch is now known as amoralej13:24
gibisean-k-mooney, kashyap: if we can always treat caps and domcaps as equivalent with the assumption of mode host-mode in case of domcaps, then I'm OK to keep a single config object for both13:51
gibimore preciesly *if we can treat the cpu part of caps and domcaps13:54
kashyapgibi: Good point.  A related thing, that you've noticed yourself, is domCaps also reports "policy"13:55
kashyapgibi: And I _don't_ think we can treat the CPU part of caps and domCaps as same ... because, here, see the diff in features from both, from my laptop - they're not equal:13:57
kashyapgibi: https://paste.opendev.org/show/bvMLvWCzjusOKOyzgsOi/14:00
kashyapThey're not equal, you see14:00
kashyap(Ignore line-1, it's accidental)14:00
gibiahh, I see, if we cannot ignore the diff as we depend on the topology for example14:01
kashyapLemme check with Jiri Denemark from libvirt14:01
gibiso I have to side with sean-k-mooney that we need to split the config model14:02
gibiand we can add helpers to move data between caps and domcaps object when needed14:02
gibikashyap: sure, more viewpoint helps14:03
sean-k-mooneythe feature listed in virsh capabliteis are the cpu feature on the host14:06
sean-k-mooneyand the model listed is the closes standard model to the host feature set14:06
kashyapYeah, as I wrote in my small primer on differences:14:06
sean-k-mooneywehn we use dom caps that actully provides what will be used by the vm when that model is selected14:06
kashyapsean-k-mooney: No, not "what will be used by the VM" - rather:14:07
sean-k-mooneyits providing the content of what we will be generted in the libvirt domain xml14:07
sean-k-mooneyif we requested host model14:07
sean-k-mooneywhich is what i ment by  "what will be used" in my previous sentence14:08
kashyapsean-k-mooney: It answers two questions: (a) whether a feature is supported by the host; and (b) whether it is supported by a given QEMU binary running on the host14:08
* kashyap --> a call; bbiab14:08
sean-k-mooneyartom: gibi  by the way i can confirm that https://review.opendev.org/c/openstack/os-brick/+/838871 fixes the unit test failures14:09
sean-k-mooneyso this backport https://review.opendev.org/q/topic:bug%252F194737014:09
sean-k-mooneyis what causes the lock path issues14:09
sean-k-mooneythe reviert is not nessisarly the correct approch14:10
sean-k-mooneybut the backport made a previously optional config option required14:10
sean-k-mooneywe neither set the config option in our test or the enviornmental variable14:10
sean-k-mooneyso strictly speaking i do no think os-brick shoudl have backported that change as written14:10
sean-k-mooneyit did not comply with satable policy since it requirs config to be updated.14:11
sean-k-mooneyelodilles:^ correct me if that is a wrong intepretation14:11
sean-k-mooneyif they provided a sane default then i think it woudl have been fine14:12
sean-k-mooneylike /run/os-brick/lock or /tmp/os-brick/lock14:12
sean-k-mooneybut in its current form its a breaking change for any deployemnt that did not set that14:13
sean-k-mooneyartom: gibi https://bugs.launchpad.net/os-brick/+bug/196979414:40
elodillessean-k-mooney: yepp, that's against stable policy and was discussed here with cinder team: https://review.opendev.org/c/openstack/releases/+/82959014:53
gibisean-k-mooney: good catch14:54
sean-k-mooneyso the option is form oslo_concurrency14:55
sean-k-mooneynot form os-brick or nova14:55
sean-k-mooneyso we can adress this in a few ways14:55
sean-k-mooneyin nova we could jsut set the env varable to somethign sane in the tox.ini14:56
sean-k-mooneythat woudl be a minimal fix we could also mock some of the calls in a fixture14:56
sean-k-mooneyapprently both nova and cinder docuemtn that this shoudl be set in the config in our install guides14:56
sean-k-mooneyso in practice all cloud should have the config option set14:56
gibiso I assume nova also uses that config option for its of file lock14:57
sean-k-mooneyelodilles: gibi  any prefernce on how to proceed.14:57
sean-k-mooneyyes14:57
sean-k-mooneywe do14:57
gibiso somewhere in the nova test env we already provide a default for that option14:57
sean-k-mooneyand os-brick is runnign in novas process space so its reading our config value14:57
sean-k-mooneygibi: that or we mock the locking code14:58
sean-k-mooneyso it does not write to disk14:58
gibiyeah14:58
gibiwe already do something so that the nova usage of that opt is covered14:58
sean-k-mooneyin this case since its failing its actully trying to create a file systm lock in the unit test14:58
sean-k-mooneywhich it shoudl not be14:58
sean-k-mooneyi woudl guess we are just missing a fixture in those tets14:58
gibiyeah that is what I think too14:59
gibiwe need to find what is missing and extend the fixture14:59
kashyapgibi: To tie up the lose end on caps vs domCaps -- the guidance from the libvirt folks is (a) no, we can't treat the caps == domCaps w/ 'host-model' mode; and (b) we should use domCaps wherever possible.15:13
kashyapgibi: So that means, we should introduce a new config object 15:13
gibikashyap: ack, make sense15:13
kashyapgibi: I might need some help on this XML parsing ... I'll take a stab15:14
kashyap... at it.15:14
gibikashyap: just ping me if you need another set of eyes15:14
kashyapWill do; thx15:14
mnaseri would appreciate some reviews on https://review.opendev.org/c/openstack/nova/+/830646 (wrt allowing addition of viommu to vms)15:59
bauzasmnaser: you're not the first one to ask for reviews on that stephenfin's old patch, will mark it as review priority for the team16:13
mnaserbauzas: cool thanks, the other one might have been ricolin but thats coming from the same side, so not sure if thats super fair ;) haha16:14
bauzasthat being said, the bp isn't validated yet16:14
bauzasand not sure stephenfin will unghost himself :)16:15
stephenfinbauzas: I think mnaser and ricolin are taking care of it now?16:15
bauzasgood question16:16
mnaseryeah it's ready (imho) from a code perspective, but if there's something else we have to do, we can take care of16:16
* mnaser has mostly been working on bug fixes and this is the most feature-esque thing in nova ever ;p16:16
bauzasmnaser: this is just a paperwork question16:17
bauzassince a new extraspec is added + the api validation16:17
bauzasthis has to be tracked correctly16:17
bauzasand we need to balance in a meeting whether we need a spec or not16:17
bauzasmnaser: that being said, I'm more than glad to welcome you as a new feature contributor ! :p16:18
bauzaswe don't have badges but I can create one for Berlin :p16:19
stephenfinI can't say if it needs a spec, but it definitely needs a blueprint and some discussion in the meeting16:19
stephenfingiven there's now an extra spec (I think?16:19
stephenfin)16:19
bauzasyeah and yeah16:20
bauzas(needs a blueprint and probably don't need a spec)16:20
bauzasstephenfin: have you just added some fancy gerrit topic for some non-existing related blueprint ?16:21
bauzasif so, you're bragging.16:21
bauzas:p16:22
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Add workaround to remove compareCPU() check on the destination  https://review.opendev.org/c/openstack/nova/+/83892616:37
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Add workaround to skip compareCPU() check on the destination  https://review.opendev.org/c/openstack/nova/+/83892616:40
sean-k-mooneymnaser: viommu is not useful without a lot of extra work17:08
sean-k-mooneystephenfin: that definetly needs a spec17:08
sean-k-mooneyviommu is trivial to enable i have implemeted that locally before17:09
sean-k-mooneymnaser: the issue is all devices will be in the same iommu group17:09
sean-k-mooneyso either you disable the memory isolation in the guest vfio-pci module17:10
sean-k-mooneyor nova has to take full contol of the pcie toplogy17:10
sean-k-mooneymnaser: stephenfin did some poc work related to this17:10
sean-k-mooneybut i dont think its complete17:11
sean-k-mooneyand it had upgrade impacts potentially17:11
stephenfinfar from complete17:11
sean-k-mooneyso it needs a spec to figure that out and more expirece playing with the poc17:11
sean-k-mooneymnaser: by the way apprently the prefomcne of the viommu is really bad with passthough devices17:12
sean-k-mooneyim not sure why but that is what the virt team told me in the past17:12
sean-k-mooneyhopefully that has changed mnaser  have you done any tests locally with livbirt?17:12
mnasersean-k-mooney: so, my knowledge level to this is that we have a company which ships pci cards which are hardware accelerators17:13
mnasersean-k-mooney: as we tried to get them to use native openstack, we figured out the extra qemu parameters they had and it included the iommu stuff17:13
mnaserSupposedly, the cards won’t work without them in PCI pass through in the guests.17:14
sean-k-mooneyack17:14
sean-k-mooneywhat driver to you use in the guest17:14
sean-k-mooneyis it vfio-pci or a normal kernel dirver17:14
sean-k-mooneyif they dont need to be in there own iommu group the stephenfin patch would enable your usecase17:14
mnaserI think you do bring up a good point that we’re adding it by default to q35 machine types, so I think making another flag to make it opt in might be good17:15
sean-k-mooneyso we could perhaps start with the simpel turn this one feature17:15
sean-k-mooneywe could possble add it by default17:15
sean-k-mooneybut what we cant do by defualt is change the pci layout to put each device in its own iommu group17:16
sean-k-mooneyat least not without a lot of testing17:16
mnaserYeah I don’t think this is what the device needs in this case17:16
sean-k-mooneyhavign a hw:viommu=on|off extra spec if simple17:16
sean-k-mooneyok stephenfin  i would be ok with a specless bluepint for just adding the extraspec/image propety for this17:17
sean-k-mooneybut if we were to do the iommu group split out that would need the spec17:17
sean-k-mooneymnaser: tldr if we want each device to be in its own iommu group we need to create a pcie expantion bridge per device. and to do that we need to assign virtual pci devivie adress to every device in the xml17:18
sean-k-mooneymnaser: so that is the bit that has upgade concerns as basically every device in the vm would change its address potentially17:19
sean-k-mooneymnaser: have you tried stephenfin's patch with that card?17:21
sean-k-mooneymnaser: im not sure im comfortable with exposeing the adress with by the way17:22
sean-k-mooneythat feel a bit too low level17:22
*** amoralej is now known as amoralej|off17:28
mnasersean-k-mooney: i haven't yet, but it's kindof a reimplementation of how they've done things with flat libvirt17:30
mnaserbut i am thinknig hw:viommu=on|off is a good idea17:31
sean-k-mooneyim about to -1 the patch17:31
mnaserin terms of controlling the address, i'm thinking since it's a flavor extra spec and not a image extra spec, it's an operator-level decision17:31
mnaserokay great17:31
sean-k-mooneyno17:31
mnaserwe'll discuss further there and ill try to poc something as well17:31
sean-k-mooneyso image extra specs are ment to be used for contoleing emulated hardware17:31
sean-k-mooneynot extra specs17:31
sean-k-mooneyso in generaly this shoudl only be exposed as an image property17:32
sean-k-mooneybut if we are to expose it in the image i would be open to having it in both17:32
sean-k-mooneyso i think we should have somehtin like  hw_viommu_modle=none|intel|smmuv3|virtio in the image to contol turing this on17:33
sean-k-mooneyand i woudl be open to also haveing  hw:viommu_modle=none|intel|smmuv3|virtio for partiy in the flavor17:33
sean-k-mooneythe adress space unless its somethign we are going to schdule on i am not conviced we shoudl expose as a configurable17:33
mnasersean-k-mooney: you know this stuff better than we do, as long as at the end of the day, we can get the xml to show up on those virt guests, we're good to go17:35
mnaserbut i agree on the image extra specs controlling hw_viommu_module too17:35
sean-k-mooneygoing forwad i think virtio is what we will want but that is very new17:36
sean-k-mooneyintel only works on x86 and smmuv3 only works on arm 17:36
sean-k-mooneyvirtio works on both17:36
sean-k-mooneymnaser: stephenfin  comment in line18:04
sean-k-mooneyhttps://review.opendev.org/c/openstack/nova/+/83064618:04
opendevreviewDan Smith proposed openstack/nova master: DNM: Run against performance.json patch  https://review.opendev.org/c/openstack/nova/+/83893418:20
dansmithgmann: ^18:20
sean-k-mooneydansmith: what is that mesuring by the way18:20
gmanndansmith: ack18:21
dansmithsean-k-mooney: https://85682c22f75746955b38-998a6cfec97762292b03290ad6103366.ssl.cf5.rackcdn.com/837139/20/check/nova-ceph-multistore/d3c4ea2/controller/logs/performance.json18:21
dansmithsean-k-mooney: a random comparison: https://termbin.com/rls018:21
sean-k-mooney so ram usage,  db queiss and api request for differnt services?18:22
dansmithfor the moment18:22
sean-k-mooneyok what is the overall goal?18:23
dansmithideally to get some flag when a patch increases one of these values substantially18:23
sean-k-mooneyby the way we had weird behavior in a downstream case where there was very high memory usage if you disabeld swap entirely18:23
dansmithlike, c-bak uses 1GiB of ram for one tempest run, which is pretty dang high.. not sure when that happened (maybe always) but..18:24
sean-k-mooneydoes it have swap enabeld on the vm?18:24
clarkbyes we enable swap on all of our test nodes18:24
sean-k-mooneyclarkb: i think some jobs turn it back off18:24
clarkbthoug hamybe that happens at a job level but it is definitely there for devstack + tempest beacuse you OOM otherwise18:25
clarkband ya swapoff is always an option18:25
dansmithyeah we're flying really close to the sun right now18:25
sean-k-mooneydansmith: we saw nova-compute and neutron l2 agents both taking over 2G each downstream when swap was off18:25
sean-k-mooneyand it went back to normal when it was turn on18:25
dansmithnot sure why that would be, but interesting18:25
sean-k-mooney... with no swap acutlly being used18:25
clarkbone of the things that happened that led to this was the rbac work drastically increased the db queries/cost and this sort of thing could catch that?18:25
dansmithclarkb: for ironic18:26
sean-k-mooneydansmith: it seamed like when swap was disabele the virtual memory and resident memory was the same18:26
sean-k-mooneybut when it was enabel there was a large deleta between the two18:26
dansmithsean-k-mooney: well, makes sense right?18:26
dansmithmaybe that's just an artifact then.. 18:26
dansmithif you're only looking at resident, then you see the working set, but if you have no other option, then rss=total18:27
sean-k-mooneywith swap on the virt memroy was the same but the prviate resent memroy was less and no swap was being used18:27
dansmithright, but it's probably because it has overcomitted to the process and until it's CoWd it can lie with swap but not if there's no swap?18:28
sean-k-mooneyi dont realy know but it did nto appre to be paging it out to swap space18:28
dansmithright, but until it needs to it won't,18:28
dansmithbut it can lie in that case18:28
sean-k-mooneybut perhaps there was some cow sematics at play18:28
dansmithwhat I mean by cow is, you can expand your heap but until you touch a page, the kernel doesn't have to actually allocate you one18:29
sean-k-mooneyyes18:29
dansmithI dunno what python's behavior is really, other than "use a crapload of memory all the time"18:30
sean-k-mooneyyou can grow the adress space but not page it to phsyical ram18:30
sean-k-mooneybut it looked liek without swap python was always commiting18:30
dansmithright, so without swap, you don't want to overcommit because the process thinks it already has that memory18:30
dansmithbut with swap, you can lie to it and offload something else during a fault18:31
sean-k-mooneyyep so just something to be aware of18:31
sean-k-mooneyif you see big difference between jobs18:31
dansmithI think turning off swap on any of our workers would likely be a big fail18:31
sean-k-mooneywe shoudl just ensure the both have the same swap config18:31
dansmiththey do, afaik18:31
dansmithbut yeah, good to know18:31
sean-k-mooneyam i think they all have 2G by default unless the providers flavor add extra swap space18:32
sean-k-mooneysome jobs increase it to 8G18:32
clarkbnone of our providers give us swap18:32
sean-k-mooneyack18:32
clarkbwe create it in the jobs18:32
sean-k-mooneyyep 18:32
clarkbthere is a difference though. Some use swapfiles and some use swap partitions18:32
clarkbwe prefer to partition but can only do that when we have a second device18:32
sean-k-mooneyyep i am reusing the upstream swap role in my ansible stuff18:32
sean-k-mooneyso i read over it 2 weeks ago18:33
sean-k-mooneyclarkb: actully since your here is there a reason thats in openstakc-zuul-jobs i was considering porting it to zuul-jobs but just said i woudl ask18:33
sean-k-mooneyhttps://github.com/openstack/openstack-zuul-jobs/tree/master/roles/configure-swap18:34
clarkbI think because its very specific to our cloud providers and the devices they expose?18:34
sean-k-mooneyhttps://github.com/openstack/openstack-zuul-jobs/blob/master/roles/configure-swap/tasks/main.yaml18:34
clarkbyou can probably make it more generic, possibly by dropping the partitioned swap option18:34
sean-k-mooneyit has some stuff for ephemeral018:34
sean-k-mooneybut it work to just use a swap file on root18:35
clarkbalso when it was first written we didn't use dd to preallicate which was fine until linux decided to break that18:35
clarkbbut now that we use dd it should work on all the filesystems that can host a swapfile18:35
sean-k-mooneythis https://github.com/openstack/openstack-zuul-jobs/blob/master/roles/configure-swap/tasks/main.yaml#L3-L29= i think is really the only non generic part18:35
sean-k-mooneybut ya no worries18:36
clarkbspecifically swapfile on ext4 created with fallocate worked until very recently on linux. But recent changes broke that and made ext4 much more like xfs and others18:36
sean-k-mooneyhum ok18:36
sean-k-mooneywe use fallocate in nova to preallocate18:37
sean-k-mooneywhat exactly changed18:37
clarkbI'd have to go find the commit that removed fallocate to remember18:37
sean-k-mooneyi think we check for odirect18:37
clarkbnote docuemntation specifically says fallocate is fine but it isn't anymore18:37
clarkbbut it is specifically for how swapfiles are managed by the kernel18:37
sean-k-mooneyok18:38
sean-k-mooneyhttps://github.com/openstack/nova/blob/master/nova/virt/libvirt/imagebackend.py#L283-L301=18:38
sean-k-mooneywe do this but perhaps that is still ok18:38
clarkbhttps://review.opendev.org/c/openstack/openstack-zuul-jobs/+/750941 is the change there is a red hat bz link that I can't open18:39
sean-k-mooney thats bad form we are ment to make them public if we are using them in commit or upstream bugs i wonder if there is a reason that lee did not in this case18:40
sean-k-mooneyoh because it was rhel9 proably in this cases18:41
sean-k-mooneyya that was commit before cenots 9 or rhel 9 were a thing publicly18:41
sean-k-mooneyclarkb: not much in it but you shoudl be able to see it now https://bugzilla.redhat.com/show_bug.cgi?id=182711518:53
sean-k-mooneytldr fallocate -z shoud be the same as dd and result in no holes18:53
clarkbI wonder if it is any faster at that point18:54
clarkbmight be worth testing that as creating the file can be slow at time18:55
sean-k-mooneyya not really sure to be honest18:55
clarkbsean-k-mooney: https://lkml.org/lkml/2020/9/24/81018:58
clarkbmaybe it is all happy now again18:59
sean-k-mooneyi cant view the last message but the previous ones in that thread look promising at least19:00
sean-k-mooneyclarkb: if there was a performace hit we could just revert lees change. shall i open a revert and see if it works?19:07
sean-k-mooneyi guess we would need a DNM test patch agaisnt devstack to test all the distros that depending on the revert19:08
sean-k-mooneyactully we proably should talk about this in #openstack-qa isntead19:09
opendevreviewsean mooney proposed openstack/nova master: enable locking test fixture  https://review.opendev.org/c/openstack/nova/+/83894219:16
sean-k-mooneyartom: gibi ^ that shoudl fix it19:16
lyarwoodsean-k-mooney: FWIW the bug was public when I linked it, it was made private by someone else a few months later.19:19
lyarwoodsean-k-mooney: also looks like that switch for fallocate landed after my original change19:19
opendevreviewGhanshyam proposed openstack/nova master: Update python classifier as per testing runtime  https://review.opendev.org/c/openstack/nova/+/83894319:20
sean-k-mooneylyarwood: ack not blaming you by the way i assumed you had a good reason to keep it private ot that it was public when you filed it but didnt check who made it private19:20
lyarwoodYeah no issues19:21
sean-k-mooneyi was going to submit a revirt but i see were are not actully passing -z19:22
artomsean-k-mooney, that make sense, but why did we not hit more often upstream>19:22
artom?19:22
sean-k-mooneyi might try updating it to use -z and regert the size ot 819219:22
artomIn downstream CI it's 100%19:22
sean-k-mooneyartom: the upper constrait bump only happend 2 months ago19:23
sean-k-mooneyso i guess we only started seeing it recently19:23
sean-k-mooneyit could be realted to mirrors having old versions19:23
sean-k-mooneybut not really sure19:23
opendevreviewGhanshyam proposed openstack/python-novaclient master: Update python classifier as per testing runtime  https://review.opendev.org/c/openstack/python-novaclient/+/83894419:24
artomsean-k-mooney, ah, so actually, lemme test that with my downstream DNM patch19:27
sean-k-mooneythat shoudl be a clean cherry pick to wallby since i actuly did it on wallayb first and then stashed it checkout masters and applied it19:31
sean-k-mooneyso i can chery pick it upstream and then do it downstream if you like19:31
opendevreviewsean mooney proposed openstack/nova stable/yoga: enable locking test fixture  https://review.opendev.org/c/openstack/nova/+/83887719:52
sean-k-mooney... damit19:52
sean-k-mooneygerrit nolonger addes the cherry-picked form lines by default19:52
sean-k-mooneyoh i bet its because its not merged.19:53
opendevreviewsean mooney proposed openstack/nova stable/yoga: enable locking test fixture  https://review.opendev.org/c/openstack/nova/+/83887719:54
opendevreviewsean mooney proposed openstack/nova stable/xena: enable locking test fixture  https://review.opendev.org/c/openstack/nova/+/83894519:55
opendevreviewsean mooney proposed openstack/nova stable/wallaby: enable locking test fixture  https://review.opendev.org/c/openstack/nova/+/83894619:55
melwittsean-k-mooney: correct, it will add the lines only if merged20:02
clarkbsean-k-mooney: ya probably worth a revert to check it20:11
opendevreviewDan Smith proposed openstack/nova master: DNM: Run against performance.json patch  https://review.opendev.org/c/openstack/nova/+/83893421:59
opendevreviewGhanshyam proposed openstack/nova master: Update python testing as per zed cycle teting runtime  https://review.opendev.org/c/openstack/nova/+/83894322:04
opendevreviewGhanshyam proposed openstack/nova master: Update python testing as per zed cycle teting runtime  https://review.opendev.org/c/openstack/nova/+/83894322:06
*** dasm is now known as dasm|off23:57

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!