Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-10-07
15:29:46 donnyd Isn't that a cloud-init function anyways
15:30:12 sean-k-mooney it is but i have tried to resize the filesystm from inside the lxc contianer manually and you cant
15:30:24 sean-k-mooney clould init works in teh lxc continer at startup
15:30:34 sean-k-mooney so i wont be able to do it either
15:30:51 donnyd Is that because of privileges to the underlying block device?
15:30:57 sean-k-mooney yep
15:31:18 sean-k-mooney we dont allow you to acess the block deivec in /dev directly with the permission to adjust the filesystem
15:31:52 donnyd So would it be possible to say start the container privileged, let cloud-init do its thing, and then restart unpriv
15:32:08 donnyd should all happen pretty quickly
15:32:25 donnyd With the understanding that right now, this is just kinda where its at
15:32:36 sean-k-mooney maybe but it might be cleaner to just have nova invoke resize2fs on the blockdevice before starting the container
15:32:51 donnyd sure, that would work too
15:33:07 sean-k-mooney i mean nova already grows the backing image to match the flavor root_disk
15:33:18 donnyd does nova blast a filesystem down onto the block device is makes now?
15:33:49 sean-k-mooney it effectly just uses qemu-image to DD the glance iamge but the resizes the image to the flavor size
15:35:03 sean-k-mooney how this actully works depends on your image backend in nova but the glance image basicaly has a partion table and filesystm
15:35:45 sean-k-mooney and we just donwlaod it and resize it. there are optimisation of copy on write formats liek qcow or backend like ceph
15:36:03 donnyd Ok, so we would have to inspect that filesystem or require meta-datas to tell nova whats in the box. Correct?
15:37:19 sean-k-mooney am im not sure we would. the qemu tools or libguestfs likely can do this for us
15:37:32 sean-k-mooney we would jsut need to call them.
15:38:42 donnyd Is libguestfs required for all distros or just RH ones?
15:39:16 donnyd Its about the slickest tool in the world for these types of things, but not sure its a req for the .debs
15:57:46 efried stephenfin: If you still have time today, wanna polish up the "SEV is busted" patch? aspiers has been MIA, so unless something drastic happens, I want to be able to merge that tomorrow so we have a bit of breathing room for the RC.
15:58:45 stephenfin Can do
16:10:22 sean-k-mooney donnyd: libguestfs is currently used for only 1 thing in nova and that thing is kind of broken
16:10:26 efried thanks stephenfin
16:11:08 sean-k-mooney donnyd: it is used to determin the default vaules of some option like disk_bus in th elibvirt dirver based on metadata on the image and i think it can do some interspection
16:11:49 sean-k-mooney so if i could remove the depency for that i would however if we had a valid usecase of it other then that i could see keeping it as a depency
16:13:13 sean-k-mooney donnyd: part of the reason i would like to remove libguestfs is currenlty it pulls in the kernel so it makes contianer image 10s-100s of megabytes bigger as it has a full kernel in the image
16:18:23 ricolin Hi, Nova team please help to review https://review.opendev.org/#/c/686996/ that error now frequently shows on grenade-heat job
16:26:49 openstackgerrit Stephen Finucane proposed openstack/nova master: docs: Rewrite host aggregate, availability zone docs https://review.opendev.org/667133
16:32:37 openstackgerrit Stephen Finucane proposed openstack/nova stable/train: docs: Highlight the current broken state of SEV https://review.opendev.org/686414
16:32:49 stephenfin efried: mriedem_afk: ^
16:33:10 openstackgerrit Merged openstack/nova stable/rocky: Add useful error log when _determine_version_cap raises DBNotAllowed https://review.opendev.org/679449
16:33:17 openstackgerrit Merged openstack/nova master: nova-net: Migrate 'test_attach_interfaces' functional tests https://review.opendev.org/684341
16:33:24 openstackgerrit Eric Fried proposed openstack/nova master: Remove workaround for bug #1709118 https://review.opendev.org/524316
16:33:24 openstack bug 1709118 in OpenStack Compute (nova) "_ContextAuthPlugin needs a refresh" [Low,Invalid] https://launchpad.net/bugs/1709118
16:33:27 openstackgerrit Merged openstack/nova master: nova-net: Migrate 'test_simple_tenant_usage' functional tests https://review.opendev.org/684342
16:33:31 efried stephenfin: fixed pep ---^
16:33:34 openstackgerrit Merged openstack/nova master: trivial: Neutron fixture cleanup https://review.opendev.org/684343
16:34:08 efried mriedem_afk: easy ksa-ish fix: Remove workaround for bug #1709118 https://review.opendev.org/524316
16:36:17 sean-k-mooney stephenfin: im sure ill find an example quickly but we can trivaly rename config vaules while both deprecteint the old name and still supporting it right.
16:36:33 sean-k-mooney stephenfin: im pretty sure we have done that several times in the past
16:36:33 stephenfin sean-k-mooney: deprecated_name='old_name'
16:36:41 sean-k-mooney cool
16:37:12 efried stephenfin, mriedem_afk: Can/should we touch the admin guide as well? (SEV caveat)
16:37:40 sean-k-mooney im wondering if we should rename reserved_huge_pages to reserved_mem_pages give it can be used to reserve mempages of any size not just hugepages
16:37:54 stephenfin fine by me
16:38:18 efried stephenfin: yeah, there's already a section on "Impermanent Limitations"
16:38:46 efried https://docs.openstack.org/nova/latest/admin/configuration/hypervisor-kvm.html#impermanent-limitations
16:38:52 sean-k-mooney stephenfin: efried did you see my comment on what actully would be broken in the reno patch
16:38:59 efried sean-k-mooney: yeah, he fixed that part
16:39:05 efried would like you to reviow
16:39:07 efried review
16:39:15 sean-k-mooney cool im finding it now
16:39:16 efried but also want to get the admin doc updated in the same patch.
16:39:24 efried sean-k-mooney: https://review.opendev.org/686414
16:40:08 sean-k-mooney thanks i just found it
16:40:49 eandersson dking_desktop, I'll let you know if I find anything
16:41:02 sean-k-mooney am the only comment i would make is ``hw_scsi_model=virtio-scsi`` and ``hw_disk_bus=scsi`` or
16:41:04 sean-k-mooney ``hw_cdrom_bus=scsi``
16:41:15 eandersson efried, you wouldn't happen to know if there has been any recent fixes to instances getting stuck in building / scheduling?
16:41:27 sean-k-mooney should be ``hw_scsi_model=virtio-scsi`` and (``hw_disk_bus=scsi`` or ``hw_cdrom_bus=scsi``)
16:41:32 efried eandersson: depends what you mean by recent, but there have been, yes, stand by...
16:41:38 eandersson :D
16:41:42 stephenfin sean-k-mooney: how about 'and either...'?
16:41:49 stephenfin X and either Y or Z
16:41:56 stephenfin I want to use English :D
16:41:56 sean-k-mooney sure i just want to make that clear
16:41:59 stephenfin yup, agreed
16:42:00 eandersson We are seeing frequent issues with instances stuck in building... with Rocky
16:42:13 sean-k-mooney e.g. hw_cdrom_bugs=scsi wont break it by itself
16:42:17 efried eandersson: oh, I thought it was merged: https://review.opendev.org/#/c/666857/
16:42:28 efried eandersson: does that ^ match your bug/symptom?
16:42:37 efried eandersson: https://bugs.launchpad.net/nova/+bug/1833581
16:42:37 openstack Launchpad bug 1833581 in OpenStack Compute (nova) "instance stuck in BUILD state if nova-compute is restarted" [Low,In progress] - Assigned to Balazs Gibizer (balazs-gibizer)
16:42:42 eandersson Yea that sounds like it
16:43:06 eandersson Took you 30s :D
16:44:07 efried eandersson: you could hit the "affects me" button. Not sure if that sends gibi an email or not...
16:44:48 efried cdent: skimming the comments, it looks like maybe we were waiting for you to take a look at that fix --^
16:45:14 openstackgerrit Stephen Finucane proposed openstack/nova stable/train: docs: Highlight the current broken state of SEV https://review.opendev.org/686414
16:45:19 cdent hello, what? which?
16:45:55 efried cdent: https://review.opendev.org/#/c/666857/
16:46:13 efried I didn't look at why, just saw a comment from mriedem_afk to that effect
16:46:42 cdent k, thanks, reading
16:46:58 efried "Let's wait to get some more feedback from cdent on this for the vcenter case"
16:47:10 efried PS11 9/18
16:49:56 efried 0_0
16:50:35 mriedem https://data.whicdn.com/images/98587406/superthumb.jpg?t=1390970016
16:51:19 mriedem gonna get lunch and check back in a bit
16:51:51 cdent efried: left a +1
16:51:58 efried thanks cdent
16:55:29 eandersson That bug sure looks like the issue. The only thing that concerns me is that we saw this hours after compute restarts, but might just have been delayed reports.
17:06:24 efried eandersson: I was +2 at some point in the past on that fix, and mriedem was just waiting for cdent to ack (I think), so we should be able to merge it fairly soon I would think. Will have to wait until next week to start merging backports, though.
17:07:02 eandersson Sounds good - we do large rolling deployments of computes
17:07:38 eandersson takes forever
17:08:15 eandersson So knowing that it's safe to do so helps a lot. Otherwise we need to disable the apis
17:25:34 donnyd sean-k-mooney: Yea that makes sense to me
17:42:00 efried stephenfin: I really appreciate the cleanup of oldy moldy change sets, but would you mind leaving a one-line reason when you abandon please?
17:47:10 mriedem efried: i'm +2 on gibi's interrupted build fix now https://review.opendev.org/#/c/666857/

Earlier   Later