| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-04-25 | |||
| 21:38:18 | mriedem | np | |
| 21:47:28 | efried | melwitt: tinyurl pointing to the charts for the project update | |
| 21:47:35 | efried | melwitt: since there's all those links on 'em | |
| 21:48:10 | melwitt | oh, the slides. yeah, that's fine | |
| 21:49:50 | mriedem | dansmith: i think the multi-cell job has yielded it's first victim | |
| 21:50:00 | dansmith | in a good way? | |
| 21:50:08 | mriedem | if latent bugs are a good thing | |
| 21:51:37 | mriedem | apparently in this addFloatingIP code, we fetch the instance from the cell at the top and pass down the targeted context for that instance, | |
| 21:51:46 | mriedem | and if the floating IP is associated with another instance we fetch that https://github.com/openstack/nova/blob/a991980863f056323c1ee9fd6a46dbc4cb899eca/nova/network/neutronv2/api.py#L2390 | |
| 21:51:59 | mriedem | but if that instance is in another cell, bingo bango you've got yourself a 404 mister | |
| 21:52:12 | mriedem | shits the bed right here https://github.com/openstack/nova/blob/a991980863f056323c1ee9fd6a46dbc4cb899eca/nova/network/neutronv2/api.py#L2400 | |
| 21:52:38 | dansmith | mriedem: pretty sure you're over your bingo bango quota for the week | |
| 21:53:05 | mriedem | umm, actually one of the cells was down and now i have no quota limit | |
| 21:53:37 | mriedem | you get to experience that kind of dry wit and more in person next week | |
| 21:54:28 | dansmith | looking forward to the "and more" there | |
| 21:54:52 | mriedem | you'll have to pay for the full riderman experience | |
| 21:55:16 | dansmith | so...many...nws...things....to...say | |
| 21:57:32 | efried | mriedem, melwitt: As soon as I can figure out how to fixture the sdk (see https://storyboard.openstack.org/#!/story/2005475) we could actually do clouds.yaml immediately. | |
| 21:57:32 | efried | Beyond that, yes, the long-term benefit is cutting the python-*client packages out of the picture, removing that tech debt. That relationship is brittle and causes us to either fear changing it or break things when we do. | |
| 21:59:28 | efried | ...immediately. Because the sdk proxies are Adapter subclasses, so we can just swap them seamlessly for the ksa adapter's we're hacking up today. Then cut the actual client libs out at our leisure. | |
| 21:59:30 | mriedem | but then we'll have (1) old style config for cinder (2) normalized ksa config for all the other non-sdk clients we use and (3) sdk clouds.yaml thing | |
| 21:59:50 | efried | deprecation period for oslo.config stuff obviously. | |
| 21:59:52 | mriedem | seems like a configuration mess until everyting is cut over | |
| 22:00:17 | efried | And operators already have clouds.yaml for the most part, so they'll have very little to do. | |
| 22:00:50 | efried | (by "already" I mean, if they're running stein+) | |
| 22:01:03 | mriedem | "That relationship is brittle and causes us to either fear changing it or break things when we do." as the recent ironicclient stuff can tell i suppose, | |
| 22:01:08 | efried | exactly | |
| 22:01:12 | dansmith | efried: which is like nobody right? | |
| 22:01:17 | mriedem | but i imagine we'll hit a new class of weird bugs once we start using the sdk | |
| 22:01:18 | efried | though I didn't want to bring that up | |
| 22:01:28 | efried | because I broke it by introducing the ksa config... | |
| 22:01:37 | mriedem | i'm not sure how much damage we get from glance/neutron/cinder client | |
| 22:02:13 | mriedem | what's being used for nova/cyborg in that series? just ksa? | |
| 22:02:16 | mriedem | or the sdk? | |
| 22:02:23 | efried | dansmith: Right, this is setting up for a future time when people are actually using the releases we're working on and we've got new jobs and don't care anymore. | |
| 22:02:35 | efried | I've been pushing Sundar to use sdk | |
| 22:03:03 | efried | I gave him the placement code for the cyborg side of his demo and set it up to use sdk | |
| 22:03:09 | mriedem | so if we did the sdk thing, cyborg integration could be the guinea pig, and then i'd think cinder since we haven't ksa'd our cinder stuff yet | |
| 22:04:14 | efried | I was going to do placement first, because it would be a transparent swap under the covers - we don't have a placementclient, so we would just set up SchedulerReportClient._client as the Connection.placement proxy (which is an Adapter subclass) instead of the raw Adapter. | |
| 22:04:25 | efried | get/put/etc primitives come along for free. | |
| 22:04:59 | efried | I PoC'd this already and it works seamlessly in tempest. It just blows up all the unit and functional tests until I figure out fixtures (see above) | |
| 22:05:21 | mriedem | ok | |
| 22:05:49 | efried | ...and then ironicclient, because we have lots of support for that from the ironic team, and dtantsur|afk is a sdk core who can help to some extent | |
| 22:06:26 | efried | and I have a guy coming on board (hi dustinc) who's trying to be involved in ironic who has started looking at it. | |
| 22:07:07 | efried | but yeah, could look into tackling cinder early, I see the benefit since it got left behind with ksa. | |
| 22:07:16 | efried | never could figure that f'in thing out. | |
| 22:15:06 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP: Correct signature on fake_get_availability_zones https://review.opendev.org/655806 | |
| 22:21:29 | mriedem | dansmith: so i'm going to rev this multi-cell job patch to disable the failing test here until there is a bug fix - while doing that, what do you think about just making the job non-voting in the check queue for a bit until we're happy with it through the ptg? | |
| 22:21:57 | dansmith | mriedem: I'm good with whatever, but it seems like non-voting to collect data is what we would normally do | |
| 22:28:02 | mriedem | ack | |
| 22:28:03 | mriedem | done | |
| 22:28:20 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add nova-multi-cell job https://review.opendev.org/655222 | |
| 22:28:21 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Enable n-novnc in nova-multi-cell job https://review.opendev.org/655711 | |
| 22:52:24 | openstackgerrit | melanie witt proposed openstack/nova master: Use futurist.ThreadPoolExecutor in scatter_gather_cells https://review.opendev.org/650172 | |
| 23:06:40 | efried | o/ see y'all tomorrow | |
| 23:41:19 | artom | ||
| 23:56:51 | rm_work | so, is it possible to run nova unit tests with tox on OSX? anyone manage this? or is it just impossible (seems that pyinotify is not compatible with osx so the env build just fails) | |
| #openstack-nova - 2019-04-26 | |||
| 01:23:34 | mriedem | dims: didn't you used to run nova unit tests in osx? ^ | |
| 01:50:53 | mnaser | I usually run them.. I haven't had troubles | |
| 01:51:10 | mnaser | i.. don't usually run the whole suite tho | |
| 01:54:50 | mriedem | tsk | |
| 02:10:52 | dims | mriedem : right as mnaser said, not the full suite ... a while ago | |
| 02:28:10 | openstackgerrit | Merged openstack/python-novaclient master: Drop py35 tests https://review.opendev.org/643873 | |
| 02:32:34 | openstackgerrit | Brin Zhang proposed openstack/nova-specs master: Change the default value of swap in the show flavor details API https://review.opendev.org/648919 | |
| 04:44:41 | openstackgerrit | melanie witt proposed openstack/nova master: Add 'path' query parameter to console access url https://review.opendev.org/649372 | |
| 05:13:35 | openstackgerrit | Merged openstack/nova master: Don't run tempest/devstack jobs on nova/test.py only changes https://review.opendev.org/655121 | |
| 06:42:30 | openstackgerrit | Brin Zhang proposed openstack/nova-specs master: Change the default value of swap in the show flavor details API https://review.opendev.org/648919 | |
| 08:28:40 | openstackgerrit | Merged openstack/nova master: Delete require_instance_exists_using_uuid https://review.opendev.org/650994 | |
| 08:33:58 | slaweq | hi nova developers :) | |
| 08:34:25 | slaweq | can someone familiar with nova-metadata take a look at my comment in https://bugs.launchpad.net/neutron/+bug/1821912/comments/12 ? | |
| 08:34:26 | openstack | Launchpad bug 1821912 in neutron "intermittent ssh failures in various scenario tests" [High,Confirmed] | |
| 08:34:45 | gibi | efried: thanks for the onboarding slide update | |
| 08:35:08 | slaweq | I'm not sure what's exectly happend (or not happend) on nova's side there so any help would be useful | |
| 08:35:32 | slaweq | gibi: efried: if You would have some time, maybe You can take a look at ^ :) | |
| 08:35:34 | slaweq | thx in advance | |
| 08:47:22 | openstackgerrit | ya.wang proposed openstack/nova-specs master: Add spec for expose-auto-converge-post-copy https://review.opendev.org/651681 | |
| 08:48:14 | gibi | slaweq: hi! looking... (but not super familiar with the metadata part of nova) | |
| 08:48:25 | slaweq | gibi: thx a lot | |
| 09:08:14 | openstackgerrit | Stephen Finucane proposed openstack/nova master: test_rpc: "deepcopy" global options https://review.opendev.org/655843 | |
| 09:20:15 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Prevent "qbr" Linux Bridge from replying to ARP messages https://review.opendev.org/655332 | |
| 09:26:57 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Always disconnect volumes after libvirtError exceptions https://review.opendev.org/655712 | |
| 09:27:18 | lyarwood | mdbooth: ^ and the one below it might interest you btw, simple bugfixes. | |
| 09:29:32 | kashyap | aspiers: Hi, when you're about, some more design comments upcoming for the getDomCapabilities() patch | |
| 09:30:30 | kashyap | aspiers: Talking to libvirt folks, they strongly suggest to not cache the capabilities at all. I will write some details and possible suggestions in the patch. | |
| 09:33:51 | aspiers | kashyap: ok | |
| 09:34:21 | kashyap | aspiers: In short, if you want to start thinking about it: we're suggested to query domCapabilities once at the beginning of building domain XML process, but then forget them right after | |
| 09:34:37 | kashyap | I asked the libvirt dev to also comment in this bug: https://bugzilla.redhat.com/show_bug.cgi?id=1683471 | |
| 09:34:39 | openstack | bugzilla.redhat.com bug 1683471 in libvirt "getDomainCapabilities claims SEV is supported for pc-i440fx-1.4 machine type" [Unspecified,Closed: notabug] - Assigned to libvirt-maint | |
| 09:34:59 | aspiers | kashyap: but this is the point I already made in the comment - that approach totally fails in the OpenStack world | |
| 09:35:13 | kashyap | In the doc string, right? | |
| 09:35:21 | aspiers | kashyap: because we need to know stuff before placement time, not just at launch time | |
| 09:35:54 | aspiers | kashyap: https://review.opendev.org/#/c/655268/2/nova/virt/libvirt/host.py@690 | |
| 09:36:17 | aspiers | I suspect the libvirt folks were only thinking of the single hypervisor paradigm when they designed this | |
| 09:36:19 | kashyap | aspiers: Right, I'm still thinking, how to resolve that | |
| 09:37:13 | aspiers | but anyway, what is the problem with memoizing? it's not caching to disk | |
| 09:37:22 | aspiers | which caps would actually change? | |
| 09:39:36 | kashyap | aspiers: For example, when QEMU is updated, you get new features | |
| 09:39:50 | kashyap | (And as you know domCapabilities are machine type dependent.) | |
| 09:40:27 | aspiers | Didn't we already talk about this? When qemu is updated, nova-compute would be restarted, so the memoization vanishes | |
| 09:40:58 | kashyap | aspiers: Meanwhile: https://bugzilla.redhat.com/show_bug.cgi?id=1683471#c11 | |