Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2023-02-16
19:27:04 sean-k-mooney mnaser: i tought there was a devstack job for this yes gmann would know more
19:27:22 mnaser i mean enabling it in nova works fine, so nova is happy, but enabling in neutron makes it un happy
19:27:39 mnaser so it could be a neutron issue so i dont know if the devstack job enables it for all or just for specific services
19:27:42 sean-k-mooney right but i tought we had it enabled for both
19:28:02 sean-k-mooney ya its a good question im not sure either
19:29:30 mnaser https://github.com/openstack/nova/blob/master/.zuul.yaml#L678-L684
19:29:31 mnaser wonder if its that
19:33:34 sean-k-mooney ya that should have it enabled for those 4 services
19:33:59 sean-k-mooney mnaser: https://zuul.openstack.org/builds?job_name=tempest-integrated-compute-enforce-scope-new-defaults&skip=0
19:34:04 sean-k-mooney it looks pretty green too
19:34:50 sean-k-mooney welll there is at least some green menaing it should work in general but im not sure how much is covered by that
19:37:22 mnaser sean-k-mooney: i wonder if we are getting hit by this since its not in zed yet - https://github.com/openstack/neutron/commit/6d8ada0ac93beed05b45adb9582c3ef23bef49d2
19:37:36 mnaser and the test that failed was actually as an admin
19:38:09 sean-k-mooney oh your trying to do this in zed
19:38:26 sean-k-mooney ya ok we only enabled it by defualt this cycle
19:39:36 sean-k-mooney mnaser: im not sure that was planned ot be backported
19:40:51 sean-k-mooney i would ask the neutrnon folk to backport it if you intend to enable it
19:41:06 sean-k-mooney its kind fo feature ish
19:41:58 mnaser sean-k-mooney: yeah i guess one could argue its a bug too
19:42:37 sean-k-mooney its because of the pivort that happend at the yoga fourm/ptg
19:42:53 sean-k-mooney when we deiced to revert a lot of the work and remove the use of scopes form most apis
19:43:22 sean-k-mooney under the orginal plan admin should not be global admin
19:43:46 sean-k-mooney so they adapted to that change in zed
21:51:09 gmann mnaser: yes, you found those. during integration testing in this cycle (tempest-full-enforce-scope-new-defaults), we found few bugs in neutron and they got fixed in master
21:51:22 gmann mnaser: I think slaweq was planning to backport those to stable/zed or older if needed
21:51:47 gmann let me find those patches
21:55:24 gmann basically these three bugs https://bugs.launchpad.net/neutron/+bug/1996150 https://bugs.launchpad.net/neutron/+bug/1996836 https://bugs.launchpad.net/neutron/+bug/1997089
21:57:58 gmann mnaser: pinged about these in neutron channel. I was in impressions that they wee backported already
22:06:58 gmann mnaser: glad to know it worked fine for nova. did you enable scope and new defaults both or just new defaults ?
23:13:41 mnaser we enabled both gmann !
23:17:52 gmann ok
#openstack-nova - 2023-02-17
01:45:43 opendevreview Merged openstack/nova master: libvirt: Add configuration options to set SPICE compression settings https://review.opendev.org/c/openstack/nova/+/828675
03:07:56 opendevreview melanie witt proposed openstack/nova master: doc: Add details about the behavior of server delete https://review.opendev.org/c/openstack/nova/+/874188
07:10:13 opendevreview Amit Uniyal proposed openstack/nova master: Added context manager for instance lock https://review.opendev.org/c/openstack/nova/+/873648
07:58:14 opendevreview Amit Uniyal proposed openstack/nova master: Added context manager for instance lock https://review.opendev.org/c/openstack/nova/+/873648
08:32:25 bauzas brightning new day, and brighting new CI failure \o/
08:46:50 opendevreview Amit Uniyal proposed openstack/nova master: Remove "see nova-manage.log" str from console msg https://review.opendev.org/c/openstack/nova/+/874206
09:02:42 opendevreview Nobuhiro MIKI proposed openstack/nova master: libvirt: Add 'COMPUTE_ADDRESS_SPACE_*' traits support https://review.opendev.org/c/openstack/nova/+/873221
10:18:57 samuelkunkel[m] this stage we use an image without uefi boot.
10:18:57 samuelkunkel[m] During the the execution of this nova will run res_req._translate_memory_encryption(request_spec.flavor, image) where the image is a just a plain objects.ImageMeta(properties=objects.ImageMetaProps())... and now nova fails (its running the function now for the second time) _check_mem_encryption_uses_uefi_image as the image does not contain useful information at all and therefore we never can create an instance as nova assumes in
10:18:57 samuelkunkel[m] Good morning, I have a question regarding AMD SEV. I am running into an issue where the the first check of hardware.get_mem_encryption_constraint is working properly (machine type correct, uefi boot check etc). further down the line nova runs libvirt_utils.get_flags_by_flavor_specs. In this function we try to fetch the ResourceRequest via scheduler_utils. At this state we only have the information about the flavor. Not the image.
10:19:06 samuelkunkel[m] Am I missing something here?
10:25:57 samuelkunkel[m] If I adjust the conditions in _check_mem_encryption_uses_uefi_image (basically if the image is just plain and does not contain any information at all) I just return (basically mocking that we use uefi here)
10:26:02 samuelkunkel[m] then it works properly
10:33:31 bauzas gibi: I may have spotted some eventlet threading issue https://paste.opendev.org/show/bFU6z82FyLG12FKMJzdN/
10:40:57 gibi bauzas: ack, I have some backlog work through so I haven't looked at the result from that yet. If you can identify the leaking test based on the new log the you can run that tests multiple times in the same executor by duplicating the test case to see if it helps reproducing locally
10:41:36 bauzas gibi: for the moment, we don't have a lot of issues with it
10:41:49 bauzas gibi: so if I have time, yeah I'll try to reproduce it
10:42:23 bauzas gibi: (it was just for telling it for you ;) )
10:42:41 gibi thanks
10:56:29 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Test aborting queued live migration https://review.opendev.org/c/openstack/nova/+/873578
10:58:17 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Test aborting queued live migration https://review.opendev.org/c/openstack/nova/+/873578
11:01:03 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Add functional tests to reproduce bug #1960412 https://review.opendev.org/c/openstack/nova/+/873579
11:52:05 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Test aborting queued live migration https://review.opendev.org/c/openstack/nova/+/873578
12:13:15 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Test aborting queued live migration https://review.opendev.org/c/openstack/nova/+/873578
12:28:24 opendevreview Alexey Stupnikov proposed openstack/nova stable/train: Test aborting queued live migration https://review.opendev.org/c/openstack/nova/+/873578
12:32:39 artom bauzas, actually let me ask here - do we have an etherpad for Bobcat already? I want to have a cross-project with the Manilla folks about what we need for Uggla's virtiofs attach feature
12:32:55 artom *Bobcat PTG, I should say
12:33:13 bauzas artom: yes we have
12:33:31 bauzas https://etherpad.opendev.org/p/nova-bobcat-ptg
12:34:53 artom Cheers!
12:35:07 bauzas Manila* dude
12:35:13 bauzas not the town :p
12:35:43 bauzas https://docs.openstack.org/manila/latest/
12:48:26 bauzas wtf, are we using cirros-0.6.1 in nova-ceph-multistore jobs ?
12:49:12 bauzas hmm no
12:50:03 bauzas hmmm, yes
12:50:09 bauzas 2023-02-17 09:45:15.491046 | controller | === cirros: current=0.6.1 uptime=103.58 ===
12:50:13 bauzas https://f89b63837ed61d9739ac-76da2058f7382f685ecdff725a6049b3.ssl.cf1.rackcdn.com/821228/7/check/nova-ceph-multistore/6617ac9/job-output.txt
12:50:21 bauzas despite the image being 0.5.2
12:56:23 kashyap bauzas: So you're saying the image is 0.5.2, but it logs it as 0.6.1?
13:22:33 bauzas kashyap: the guest console, yes
13:22:59 bauzas and it uses the dhcpcd client for getting the lease
13:24:51 bauzas "use Ubuntu 22.04 (5.15) kernels for aarch64, arm, ppc64le, x86_64"
13:24:55 opendevreview Samuel Kunkel proposed openstack/nova master: fix: handle missing img properties https://review.opendev.org/c/openstack/nova/+/874248
13:25:15 bauzas compared to 0.5.0 "Linux kernel 5.3.0-26.28~18.04.1 for aarch64, arm, i386, ppc64le, x86_64 (Ubuntu 18.04 HWE)"
13:25:52 samuelkunkel[m] please have a look and provide me some feedback if the approach is correct. With that "fix" everything works properly for deploying AMD SEV
13:25:59 bauzas so, yeah the guest that was booted was using 0.6.1
13:26:02 bauzas 2023-02-17 09:45:15.450272 | controller | [ 0.000000] Linux version 5.15.0-48-generic (buildd@lcy02-amd64-080) (gcc (Ubuntu 11.2.0-19ubuntu1) 11.2.0, GNU ld (GNU Binutils for Ubuntu) 2.38) #54-Ubuntu SMP Fri Aug 26 13:26:29 UTC 2022 (Ubuntu 5.15.0-48.54-generic 5.15.53)
13:26:36 bauzas samuelkunkel[m]: I'm not really knowing about AMD SEV but I'll try to review your change
13:29:15 kashyap bauzas: Hmm, thx for the kernel comparision.
13:30:37 samuelkunkel[m] thanks bauzas I am also not that familiar but I tried to describe the implications
13:31:54 kashyap bauzas: I see the code from samuelkunkel[m] is not in core SEV functionality itself, but in sched/utils
13:32:27 kashyap samuelkunkel[m]: What triggerred this change? Would be useful if you could write-up a small bug report / tracker
13:34:05 samuelkunkel[m] I can write up a bug report
13:34:46 samuelkunkel[m] yes, its with in the scheduler utils but its triggert by the AMD SEV implementation
13:35:15 samuelkunkel[m] This fix "could" (in a somewhat similar way) also be moved to the memory encryption section
13:35:29 samuelkunkel[m] e.g. directly into _check_mem_encryption_uses_uefi_image where its triggered
13:37:53 kashyap samuelkunkel[m]: I see, then what made you move it scheduler/utils? If it could be dealt with in nova/virt/hardware.py)
13:38:18 kashyap Thanks for writing it! (And link it in the commit message on your next respin, plese)
13:41:18 opendevreview Kashyap Chamarthy proposed openstack/nova stable/wallaby: Add a workaround to skip hypervisor version check on LM https://review.opendev.org/c/openstack/nova/+/851206
13:41:19 opendevreview Kashyap Chamarthy proposed openstack/nova stable/wallaby: libvirt: At start-up rework compareCPU() usage with a workaround https://review.opendev.org/c/openstack/nova/+/872402
13:59:07 opendevreview Andre Aranha proposed openstack/nova stable/yoga: [stable-only] Test setting the nova job to centos-9-stream https://review.opendev.org/c/openstack/nova/+/860087
14:00:15 opendevreview Andre Aranha proposed openstack/nova stable/yoga: [stable-only] Test setting the nova job to centos-9-stream https://review.opendev.org/c/openstack/nova/+/860087
14:00:58 opendevreview Andre Aranha proposed openstack/nova stable/yoga: [stable-only] Test setting the nova job to centos-9-stream https://review.opendev.org/c/openstack/nova/+/860087
14:27:06 samuelkunkel[m] https://bugs.launchpad.net/nova/+bug/2007697
14:27:34 samuelkunkel[m] I hope I got everything. I will try to move the fix into the virt/hardware.py. It should not a big issue to just exclude the uefi check if no image informations are present at all
15:22:00 bauzas was fucking close to get merged https://review.opendev.org/c/openstack/nova/+/872413 :cries:
15:23:37 opendevreview Samuel Kunkel proposed openstack/nova master: fix: handle missing img properties https://review.opendev.org/c/openstack/nova/+/874248
15:24:25 gibi bauzas: I will look at the etherpad during the weekend and try to recheck-in thigns we already approved

Earlier   Later