Uggla | o/ | 07:35 |
---|---|---|
opendevreview | Elod Illes proposed openstack/nova stable/train: [stable-only] Use Tempest's run upper constraints from devstack https://review.opendev.org/c/openstack/nova/+/842813 | 07:35 |
gibi | good morning | 07:53 |
*** elodilles is now known as elodilles_afk | 08:31 | |
kashyap | melwitt: Thanks for that workaround for the 'query-migrate' call: https://review.opendev.org/c/openstack/nova/+/842687 :) | 08:31 |
opendevreview | Balazs Gibizer proposed openstack/nova master: Revert "Enable live_migration_events in nova-ovs-hybrid-plug" https://review.opendev.org/c/openstack/nova/+/842932 | 08:39 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/xena: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842922 | 09:10 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/xena: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842923 | 09:10 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/wallaby: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842925 | 09:11 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/wallaby: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842926 | 09:11 |
*** ministry is now known as __ministry | 09:48 | |
opendevreview | Balazs Gibizer proposed openstack/nova stable/victoria: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842952 | 09:49 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/victoria: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842953 | 09:49 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/victoria: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842952 | 09:58 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/victoria: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842953 | 09:58 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/ussuri: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842955 | 10:03 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/ussuri: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842956 | 10:03 |
opendevreview | Merged openstack/nova-specs master: Re-propose spec for ephemeral storage encryption https://review.opendev.org/c/openstack/nova-specs/+/835877 | 10:35 |
opendevreview | Merged openstack/nova-specs master: Re-propose spec for ephemeral encryption for libvirt https://review.opendev.org/c/openstack/nova-specs/+/836075 | 10:37 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/train: Reproducer unit test for bug 1934094 https://review.opendev.org/c/openstack/nova/+/842962 | 11:23 |
opendevreview | Balazs Gibizer proposed openstack/nova stable/train: Fix instance's image_ref lost on failed unshelving https://review.opendev.org/c/openstack/nova/+/842963 | 11:23 |
songwenping | bauzas: morning, have you used Tesla A100 card, how to use it binding to vm, is there any docs? | 11:43 |
sean-k-mooney | songwenping: i dont think we have them installed in our lab yet. we were goingt ot be adding some | 11:50 |
sean-k-mooney | songwenping: i think its more or less the same as the previous generation | 11:50 |
songwenping | sean-k-mooney: passthrough is same as the previous generation | 11:51 |
songwenping | but vgpu usage is different. | 11:51 |
sean-k-mooney | vGPU can be too | 11:51 |
sean-k-mooney | it depense on if you enabel mig or not | 11:51 |
sean-k-mooney | form an openstack point of view even with mig enabled the nova config is baically the same | 11:51 |
sean-k-mooney | we still use mdevs for vGPU in mig mode | 11:52 |
songwenping | i have enabled mig, but i cannot find the vgpu types. | 11:52 |
sean-k-mooney | but the mdev is moved to the mig vf | 11:52 |
sean-k-mooney | songwenping: are you checking the PF or the VFs | 11:52 |
songwenping | i exec the sirov-manage, but it has no effect. | 11:53 |
sean-k-mooney | i have not used it but nvidia provide https://github.com/NVIDIA/mig-parted | 11:53 |
sean-k-mooney | to configre the a100 | 11:53 |
sean-k-mooney | this would appear to be the set of supported profiles https://docs.nvidia.com/datacenter/tesla/mig-user-guide/index.html#a100-profiles | 11:55 |
sean-k-mooney | "sudo nvidia-smi mig -lgip" shoudl list the profiles for you | 11:56 |
songwenping | we also need create mdev then attach to the guest? | 11:56 |
sean-k-mooney | the mdevs i think will still be created by nova | 11:57 |
sean-k-mooney | but you need to lookup the supproted mdev type in /sys | 11:57 |
sean-k-mooney | so with mig mode there are 2 steps | 11:58 |
songwenping | yeah, to generate uuid under devices dir by nova | 11:58 |
sean-k-mooney | partition the gpu wiht mig using the mig profiles | 11:58 |
sean-k-mooney | then lookup the mig vf pci adress and mdev type and update nova.conf with that | 11:58 |
songwenping | ok, i'll try to check. | 11:59 |
sean-k-mooney | then nova will determin the numer of that mdev type that can be created and create an inventory of thaty type inplacement. when you schedule a vm we shoudl create the mdev automaticlly | 11:59 |
songwenping | right, thanks seak-k-mooney. | 12:00 |
songwenping | sorry, sean-k-mooney. | 12:01 |
sean-k-mooney | hehe given i have missspelled my own name in code comment no need to appologies for a typo :) | 12:02 |
sean-k-mooney | i was just lookign for our downstream bz for a100 but i cant find it readialy | 12:02 |
sean-k-mooney | i think bauzas wrote up how to do this at some point. they had a load of an 100 to test breifly but we dont have one dedicated for our use currently | 12:03 |
sean-k-mooney | ah i found it but its private | 12:05 |
songwenping | look forward to seeing soon. | 12:08 |
sean-k-mooney | it looks like they were using /usr/lib/nvidia/sriov-manage -e 0000:04:00.0 | 12:09 |
sean-k-mooney | to allocate teh VFs | 12:09 |
songwenping | i also use this command | 12:09 |
sean-k-mooney | and then if you used ll /sys/class/mdev_bus/* | 12:09 |
sean-k-mooney | you woudl see the vfs | 12:09 |
sean-k-mooney | are mdev capable | 12:09 |
sean-k-mooney | they saw the same behvior with both t4(with updated firmware) and a100 | 12:10 |
songwenping | but get nothing, maybe my env has something problem | 12:10 |
sean-k-mooney | then they create the mig instance "nvidia-smi mig -cgi 2g.10gb,2g.10gb,2g.10gb -C" | 12:10 |
sean-k-mooney | and the vf had inventories | 12:10 |
sean-k-mooney | cat /sys/class/mdev_bus/0000\:04\:01.0/mdev_supported_types/nvidia-475/available_instances | 12:10 |
sean-k-mooney | so they added | 12:11 |
sean-k-mooney | [devices] | 12:11 |
sean-k-mooney | enabled_vgpu_types = nvidia-475,nvidia-1 | 12:11 |
sean-k-mooney | [vgpu_nvidia-475] | 12:11 |
sean-k-mooney | device_addresses = 0000:04:00.4,0000:04:00.5,0000:04:00.6 | 12:11 |
sean-k-mooney | [vgpu_nvidia-1] | 12:11 |
sean-k-mooney | device_addresses = 0000:00:00.1 | 12:11 |
sean-k-mooney | to the nova.conf | 12:11 |
sean-k-mooney | songwenping: the steps bauzas did were | 12:12 |
sean-k-mooney | nvidia-smi -i 0 -mig 1 | 12:12 |
songwenping | thanks sean-k-mooney | 12:12 |
sean-k-mooney | to enable mig | 12:12 |
sean-k-mooney | which they checked with | 12:12 |
sean-k-mooney | nvidia-smi -i 0 --query-gpu=pci.bus_id,mig.mode.current --format=csv | 12:13 |
sean-k-mooney | then "/usr/lib/nvidia/sriov-manage -e 00:04:0000.0" to enable the vfs | 12:13 |
sean-k-mooney | and then you can use " ll /sys/class/mdev_bus/" | 12:13 |
sean-k-mooney | to verify that hte device are capable of hosting mdevs | 12:13 |
sean-k-mooney | then they configugred vgpu instance per mig instance | 12:14 |
sean-k-mooney | nvidia-smi mig -cgi 2g.10gb,2g.10gb,2g.10gb -C | 12:14 |
sean-k-mooney | and after that they had 1 mdev inventory per vf | 12:14 |
sean-k-mooney | which can be checked with "cat /sys/class/mdev_bus/0000\:04\:01.0/mdev_supported_types/nvidia-475/available_instances" | 12:15 |
sean-k-mooney | songwenping: hopefully that helps | 12:15 |
songwenping | thanks, this is useful. | 12:15 |
*** elodilles_afk is now known as elodilles | 12:16 | |
sean-k-mooney | our downstream docs bz to write this up is still pending | 12:16 |
sean-k-mooney | dansmith: those keystone select number really are an order of magnitude higher then everything else https://paste.opendev.org/show/bsG9exlvGqaNZrbhEnJp/ | 12:34 |
gibi | melwitt: can I help somehow to convince you to +2 https://review.opendev.org/c/openstack/nova-specs/+/791047 ? :) | 12:59 |
sean-k-mooney | gibi: she may have been leaving that to stephenfin to +2w but if melwitt is happy withthe proposal i think stephenfin will be happy that you have incoperated there feedback with regards to the dynmaic alloctaiton fo VF/PFs | 13:03 |
gibi | ahh OK. | 13:05 |
gibi | stephenfin: if you have time, I think https://review.opendev.org/c/openstack/nova-specs/+/791047 is ready for a +2+A :) | 13:05 |
* sean-k-mooney agrees for what its worth but may need to ensure they are sitting down if that actully merges | 13:06 | |
gibi | I'm not sure I understand what the sitting down part means ^^ | 13:08 |
sean-k-mooney | preparing my self for the shock of us agreeing to actully track pci device in placement after like what 5 years | 13:09 |
gibi | ahh :d | 13:09 |
gibi | :D | 13:09 |
sean-k-mooney | placement was still in nova wehn we started this converastion and nested resouce providers did not exist | 13:09 |
gibi | yeah it is a long journey | 13:10 |
gibi | and it is not over yet :) | 13:10 |
opendevreview | ribaudr proposed openstack/nova-specs master: libvirt: Allow Manila shares to be directly attached to instances https://review.opendev.org/c/openstack/nova-specs/+/833669 | 13:32 |
dansmith | sean-k-mooney: I know.. something has to be wrong right? | 13:49 |
sean-k-mooney | presumable. i would expect keysone to be higher then the rest but not 10x | 13:50 |
sean-k-mooney | and that 10x neutron which is also higher then the rest | 13:51 |
dansmith | right | 13:51 |
dansmith | neutron goes over keystone quickly during tempest tho | 13:52 |
sean-k-mooney | dansmith: while your here we use 2 allocation for live migration correct? one is held by the instnace an the other by the migration uuid? | 14:11 |
dansmith | yeah | 14:11 |
sean-k-mooney | do you remember when that change was made it was 5 or more cycles ago right | 14:11 |
sean-k-mooney | as in its not recent | 14:11 |
sean-k-mooney | https://bugs.launchpad.net/nova/+bug/1975490 just in the context of that | 14:11 |
dansmith | oh jeez, no idea, but I'm sure it was 5+ ago | 14:11 |
sean-k-mooney | right it was well before train with the new vcpu code | 14:12 |
sean-k-mooney | they did not say what version of openstack they were using but im guessign it was quite an old version if they are seeing placment check the source allocations like they describe | 14:13 |
sean-k-mooney | i dont think any supported branch including the ones in EM still use a signel allocation for both hosts | 14:14 |
dansmith | yeah I'm sure not | 14:15 |
sean-k-mooney | we will see if they respond i guess | 14:15 |
opendevreview | ribaudr proposed openstack/nova-specs master: Allow unshelve to a specific host https://review.opendev.org/c/openstack/nova-specs/+/831506 | 14:30 |
*** mfo is now known as Guest67 | 14:47 | |
*** mfo_ is now known as mfo | 14:47 | |
kashyap | melwitt: When you're about, can you review this, please: https://review.opendev.org/c/openstack/nova/+/838926 (It already has a +2 from gibi) | 15:14 |
kashyap | (And it is successfully tested by a Red Hat contributor downstream) | 15:15 |
kashyap | bauzas: --^ | 15:18 |
*** beekneemech is now known as bnemec | 15:39 | |
melwitt | bauzas, elodilles, gmann: I started an etherpad for stable branch ci here. will fill it in with more info https://etherpad.opendev.org/p/nova-stable-branch-ci | 15:42 |
melwitt | kashyap: thanks for noticing :) | 15:43 |
melwitt | gibi, sean-k-mooney: yeah I thought stephenfin may want to look at the current spec bc he had a lot of feedback in previous PS | 15:44 |
stephenfin | oh, I'm looking at it at the moment | 15:44 |
kashyap | melwitt: I recall debugging it with you in the past; thanks for bringing it to its logical conclusion | 15:44 |
kashyap | Looks good to me to workaround it. (Also, the CI failure seems unrelated) | 15:45 |
melwitt | stephenfin: great :) | 15:45 |
melwitt | kashyap: will also review your patch | 15:46 |
kashyap | melwitt: It's also a workaround :D | 15:46 |
melwitt | haha nice | 15:46 |
elodilles | melwitt: thanks \o/ it was on my TODO as well o:) | 15:46 |
kashyap | melwitt: But it is functionally tested by a Red Hat support eng downstream | 15:47 |
kashyap | (And it works) | 15:48 |
gmann | melwitt: ack, thanks. | 15:48 |
melwitt | kashyap: ack | 15:48 |
kashyap | Thank you | 15:49 |
elodilles | melwitt: i proposed quite many bandit capping patches in the past, but did not notice that placement needs it :-o ( https://review.opendev.org/q/topic:cap-bandit-1.6.2 ) | 15:54 |
melwitt | elodilles: yeah I remembered that being a thing and I think I found your patches for other projects. placement had been working ok for awhile, not actually sure when this cropped up. backports don't get proposed there often | 15:56 |
bauzas | Uggla: made a review with a French link on https://review.opendev.org/c/openstack/nova-specs/+/831506/ | 15:58 |
* bauzas did this for the first time in 9 years :) | 15:58 | |
bauzas | either way, +2d | 15:58 |
bauzas | cores, a spec seems very close to be accepted https://review.opendev.org/c/openstack/nova-specs/+/831506/4.55 | 15:58 |
bauzas | cores, a spec seems very close to be accepted https://review.opendev.org/c/openstack/nova-specs/+/831506/4..5 even | 15:58 |
* bauzas feels brave to review https://review.opendev.org/c/openstack/nova-specs/+/791047 now as he promised too much on it | 15:59 | |
elodilles | melwitt: yeah, and we have periodic-stable-jobs in placement only since xena: https://review.opendev.org/c/openstack/placement/+/775384 | 16:00 |
stephenfin | gibi: sean-k-mooney: melwitt: My only comment is on the config option. 'device_list' suggests it's a list but it's not - it's a MultiStrOpt (meaning you'd simply specify it multiple times in nova.conf) | 16:01 |
melwitt | elodilles: ah right | 16:02 |
elodilles | melwitt: we have the periodic-weekly for master, but maybe this worth a backport if we want to catch broken gates on older stable branches | 16:02 |
gmann | bauzas: FYI, i will be absent in nova meeting tomorrow as we have 'release naming meeting' at the same time - http://lists.openstack.org/pipermail/openstack-discuss/2022-May/028618.html | 16:02 |
stephenfin | We can bikeshed on the correct name for that now or later, I don't mind. So long as we don't call it a plural thing when it's not actually plural | 16:02 |
gibi | stephenfin: hm | 16:04 |
sean-k-mooney | stephenfin: well eventully we shoudl make it a list instad of a multi opt | 16:04 |
sean-k-mooney | stephenfin: the alisa support json list syntax or mutlti opt repating | 16:04 |
sean-k-mooney | stephenfin: generally it enables multipel things by the way | 16:05 |
sean-k-mooney | it supprots regex and glob mathch or you can just use vender id and product id | 16:05 |
gibi | stephenfin: we can drop the _list postfix and simply call it [pci]devices | 16:05 |
sean-k-mooney | ya that would work | 16:06 |
sean-k-mooney | gibi: stephenfin if ye want to s/device_list/devices/ im fine with that | 16:12 |
sean-k-mooney | and or do it as a followup | 16:12 |
gibi | as I see stephenfin has things against the pluralness as well | 16:12 |
gibi | anyhow I will reply in the spec and we can discuss it tomorrow over IRC | 16:13 |
sean-k-mooney | well from my point of view it is plural as it generally refers to multiple devices via the vendor/product id or the regex support | 16:13 |
sean-k-mooney | sure | 16:13 |
sean-k-mooney | elodilles: by the way i propsoed https://review.opendev.org/c/openstack/requirements/+/842991 to test if that will help with stable/ussuri | 16:14 |
sean-k-mooney | elodilles: i think that woudl pull in most of the wait for sshabel changes but certenly not all | 16:14 |
sean-k-mooney | there are some other idea we had of how to impove the stabel branch situation but that was the lowest effort quick change we cam up with | 16:15 |
sean-k-mooney | ah actully i might need older | 16:16 |
sean-k-mooney | we might be hitting py36 issues witht hat | 16:16 |
sean-k-mooney | although we should supprot 36 on yoga so proably somethign else | 16:17 |
sean-k-mooney | ah oslo_log dependices | 16:18 |
sean-k-mooney | hum why is that using upper constratis form master | 16:19 |
sean-k-mooney | https://zuul.opendev.org/t/openstack/build/1f08c1134a1540bb8e9ae836c5ba60a9/log/job-output.txt#20845 ERROR: Could not find a version that satisfies the requirement oslo.log===5.0.0 (from -c https://releases.openstack.org/constraints/upper/master | 16:19 |
bauzas | gmann: ack no worries | 16:20 |
sean-k-mooney | ok i need to also change devstack to work aroudn that | 16:21 |
gibi | dansmith: could you check the unshelve to host spec https://review.opendev.org/c/openstack/nova-specs/+/831506/ I think now it is also aligned to you comments too | 16:26 |
dansmith | oh yeah I saw this ack | 16:27 |
dansmith | oops | 16:27 |
dansmith | that was intended to be two statements to two different people, but it worked out this time :) | 16:27 |
* dansmith 's irc client has one input buffer, which persists when switching tabs :/ | 16:27 | |
sean-k-mooney | oh i have debated if that was useful in the past | 16:28 |
sean-k-mooney | i sometimes start replyint in the wrong tab if misclick | 16:28 |
sean-k-mooney | but more often then not i assuem that causes issues | 16:28 |
gibi | :) | 16:29 |
dansmith | I'm not a fan.. end up with too many "that guy is an ass okay sure buddy" things where you don't remember to hit send in a /query and then switch | 16:29 |
sean-k-mooney | https://review.opendev.org/c/openstack/requirements/+/842991 is going to fail but even using an old upper-constriats i think wont work | 16:33 |
sean-k-mooney | its been resolving for ~5 mins locally | 16:33 |
sean-k-mooney | realistically if i want to bump the tempest version i think i would need to either do this differently via devstack or bump other constraits | 16:34 |
sean-k-mooney | actully the best apptoch might be to just not use an upper-constratits file for the venv | 16:41 |
dansmith | gibi: I have one typo gripe plus bauzas | 16:44 |
dansmith | can we fix and re-ack real quick? | 16:44 |
dansmith | I can edit inline and then +2 if you'll be around to +W | 16:44 |
bauzas | dansmith: sure | 16:44 |
opendevreview | Dan Smith proposed openstack/nova-specs master: Allow unshelve to a specific host https://review.opendev.org/c/openstack/nova-specs/+/831506 | 16:45 |
bauzas | gibi: I'll have comments on the PCI spec, mostly on the config options | 16:45 |
sean-k-mooney | ok i think the best way to make this work is just to remove using upper-constraits from the venv in devstack ill try that and see if that work | 16:45 |
dansmith | bauzas: gibi done | 16:45 |
bauzas | dansmith: saw | 16:45 |
bauzas | Uggla: typey typey now | 16:46 |
Uggla | ? | 16:46 |
bauzas | Uggla: your spec got approved | 16:53 |
bauzas | which means "hands on deck now" | 16:53 |
Uggla | oh cool | 16:53 |
gibi | dansmith, bauzas: thanks | 16:59 |
sean-k-mooney | gmann: o/ hi looks like you are also tryinng to use a newer version of tempest on ussuri | 17:04 |
sean-k-mooney | https://review.opendev.org/c/openstack/devstack/+/838051 | 17:04 |
sean-k-mooney | im trying to do the same with https://review.opendev.org/c/openstack/requirements/+/842991 andhttps://review.opendev.org/c/openstack/devstack/+/843006 | 17:04 |
sean-k-mooney | ok im going to call it a day but ill take a look at those again tomorrow | 17:07 |
opendevreview | Merged openstack/nova-specs master: Allow unshelve to a specific host https://review.opendev.org/c/openstack/nova-specs/+/831506 | 17:14 |
melwitt | sean-k-mooney: have you seen this re: upper-constraints? https://review.opendev.org/c/openstack/nova/+/842813 | 17:16 |
melwitt | oh, nvm, I guess that's what you just talking about the last few messages | 17:16 |
sean-k-mooney | no i had not | 17:18 |
sean-k-mooney | but i dont think we want to do that | 17:18 |
sean-k-mooney | i dont really agree with pinnint tempest but if we do im not conviced we shoudl do it via devstack | 17:18 |
sean-k-mooney | we could but i think we need to contol that form the project side unless the QA team want to support all the brances we care about including the EM ones | 17:19 |
melwitt | it's a nova patch | 17:20 |
sean-k-mooney | i know but devstack does nto seam to use the right version | 17:20 |
sean-k-mooney | either | 17:20 |
sean-k-mooney | at least on the stable branch of the requirement repo it uses master | 17:20 |
sean-k-mooney | not the patch under test | 17:21 |
sean-k-mooney | it looks like on stable its using the checked out version of the requiremetns repo | 17:22 |
sean-k-mooney | well on train https://github.com/openstack/devstack/blob/stable/train/stackrc#L320 | 17:22 |
sean-k-mooney | that would be fine i guess if tha tis the correct branch | 17:22 |
sean-k-mooney | melwitt: https://github.com/openstack/devstack/blob/stable/ussuri/stackrc#L315= | 17:23 |
sean-k-mooney | on ussui it woudl end up using master | 17:23 |
sean-k-mooney | melwitt: so i guess that patch is fine for train | 17:23 |
sean-k-mooney | but not for ussuri | 17:23 |
melwitt | hm ok | 17:24 |
sean-k-mooney | look like gmann pinned train with https://github.com/openstack/devstack/commit/8a22f7380c7029e931fe9103320f24a223b619d1 | 17:25 |
sean-k-mooney | so i guess the train patch is fine to merge | 17:26 |
sean-k-mooney | but that would not help with ussuri | 17:26 |
gibi | bauzas: thanks for the review, replied in https://review.opendev.org/c/openstack/nova-specs/+/791047 | 17:26 |
sean-k-mooney | melwitt: i kind of would have epected the newer stable branches to be fix before train | 17:26 |
sean-k-mooney | but maybe its because train is EM already | 17:27 |
opendevreview | ribaudr proposed openstack/nova-specs master: libvirt: Allow Manila shares to be directly attached to instances https://review.opendev.org/c/openstack/nova-specs/+/833669 | 17:27 |
sean-k-mooney | melwitt: mind if we pick this up tomorrow the train nova patch is likely ok to proceed with | 17:27 |
melwitt | sean-k-mooney: it might be an oversight? gmann ^ | 17:27 |
melwitt | sean-k-mooney: sure | 17:27 |
sean-k-mooney | i would think using the locally checked out requiremetn repo would always be valid | 17:28 |
sean-k-mooney | and that instead of a stable only change we coudl have done this on master and backported | 17:28 |
sean-k-mooney | at least on the nova side | 17:28 |
sean-k-mooney | anyway got to run and pick up food for dinner | 17:29 |
melwitt | yeah, that does make sense | 17:29 |
melwitt | o/ | 17:30 |
elodilles | melwitt: about the TEMPEST_VENV thing: afaik that is intentional: tempest should run always against master in 'maintained' stable branches (that's why it is in venv) | 17:59 |
melwitt | elodilles: I replied on the review, I see what you mean, it was my misunderstanding. I see now that tempest is pinned on train but not elsewhere | 18:00 |
melwitt | thanks for the quick reply :) | 18:00 |
elodilles | melwitt: oh, sorry :) haven't seen your answer yet o:) | 18:01 |
melwitt | no worry, I posted it only a minute ago :) | 18:01 |
elodilles | :) | 18:02 |
melwitt | I should have said "I also replied on the review" | 18:02 |
elodilles | no problem :) i really wonder how these new problems crept into ussuri and train... i mean, the SSHABLE thing was not needed before, so why now? did we merge something to nova? or did it come with some other dependencies? :/ | 18:08 |
melwitt | elodilles: same. looking at a CI run from today on stable/ussuri, one thing it's failing on the same oslo.log dep issue as on stable/train https://zuul.opendev.org/t/openstack/build/571459b47ccb4ed1854f61ce99597eca | 18:12 |
melwitt | ERROR: full: could not install deps [-chttps://releases.openstack.org/constraints/upper/master, -r/opt/stack/tempest/requirements.txt]; v = InvocationError('/opt/stack/tempest/.tox/tempest/bin/pip install -chttps://releases.openstack.org/constraints/upper/master -r/opt/stack/tempest/requirements.txt', 1) | 18:14 |
elodilles | melwitt: oslo.log 5.0.0 dropped py36 & py37 | 18:29 |
elodilles | so yes it seems it is used from master's upper-constraints.txt :S | 18:30 |
elodilles | but on ussuri we have zuulv3. so you are right, we need a fix for ussuri then, too :S (though it will be probably a different fix, because the hook does not exist there anymore) | 18:33 |
melwitt | yeah :/ | 18:35 |
melwitt | sean-k-mooney has been looking at it | 18:35 |
elodilles | i guess from zed till victoria now things are running with py38 and now that during zed some project dropped py36 support we need to pin things where still py36 is used only | 18:46 |
elodilles | i.e. on ussuri and train | 18:46 |
sean-k-mooney | elodilles: the sshable thing is needed partly because of a qemu change and partly because it was always a race | 20:24 |
sean-k-mooney | elodilles: it was previoulsy undefined behaivor to retry detach | 20:24 |
sean-k-mooney | we have always had races in those tests too with the kernel attachiting the device so it was technially always need but it was not an issue in the past | 20:25 |
sean-k-mooney | i dont know if canonical have backproted the change or centos have in the stable brances to qemu | 20:25 |
sean-k-mooney | but it could jsut be that the ci is runing slightly slower and now we are lossing rahter then wining the race | 20:26 |
sean-k-mooney | elodilles: using master requirement with master tempest makes sesnse but on any branch where we pin tempest we have to also pinn the upper constraits | 20:27 |
sean-k-mooney | we also need to deal with py36 by either using py38 to run tempest | 20:27 |
sean-k-mooney | on stable ussuri or using older requirements to work around oslo droping support | 20:28 |
sean-k-mooney | basiclaly not that master does not use py36 we cant use master in any job the uses py36 for the upper constraits so we need to cap to yoga at the latest | 20:30 |
opendevreview | Merged openstack/nova stable/train: [stable-only] Use Tempest's run upper constraints from devstack https://review.opendev.org/c/openstack/nova/+/842813 | 21:17 |
gmann | sean-k-mooney: I did not get about your 'unpined' things. For all EM, we pin Tempest in devstack and so does compatible upper constraints to use. for ussuri, I am trying with tempest 30 but I have not finished that yet. | 23:43 |
gmann | sean-k-mooney: note, project can always override the pin tempest version and constraints via devstack variable in job so projects side choice still available | 23:43 |
gmann | I am not clear what exactly we miss to pin the Tempest for EM branch. All good there right? | 23:45 |
melwitt | gmann: I put the ussuri issue on the etherpad https://etherpad.opendev.org/p/nova-stable-branch-ci basically it's failing the same way train was, failing to install oslo.log version coming from upper-constraints master | 23:50 |
gmann | melwitt: thanks. I started working on this but then forget to figure out the failure and release - https://review.opendev.org/q/topic:ussuri-last | 23:54 |
gmann | I will work on that | 23:54 |
melwitt | ah k cool | 23:58 |
gmann | due to some constraint mismatch in Tempest master and stable/ussuri, it is little complicated and I am trying to see what combination of Tempest and constraint will work | 23:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!