| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-04-20 | |||
| 13:45:06 | fried_rice | finucannot: You can't, like, jump into the middle of a function and start mocking, if that's what you're asking. | |
| 13:46:40 | dansmith | sean-k-mooney[m]: around? | |
| 13:47:01 | finucannot | fried_rice: Ah, fair. I was hoping for a 'passthrough=True' flag. Doesn't seem it does that | |
| 13:47:09 | finucannot | Guess it's kind of the opposite of what you'd want | |
| 13:47:29 | fried_rice | finucannot: yeah, because it would have to pass through conditionally, and you'd need some way to specify that condition. | |
| 13:47:40 | fried_rice | finucannot: Otherwise, if you wanted it to pass through always, you would just... not mock it. | |
| 13:51:55 | fried_rice | finucannot: I'm ready to give up trying to build docs locally. tox -re placement-api-ref is just sitting forever at | |
| 13:51:55 | fried_rice | placement-api-ref installdeps: -r/home/efried/Neo/nova/doc/requirements.txt | |
| 13:52:33 | finucannot | fried_rice: Just use 'tox -e docs'? | |
| 13:52:39 | fried_rice | finucannot: same | |
| 13:52:45 | fried_rice | tried that earlier | |
| 13:52:53 | finucannot | Hmm, that's a problem on your end, I'd imagine | |
| 13:52:57 | finucannot | Working a-ok here | |
| 13:53:07 | fried_rice | it spun, seemingly doing nothing, for a good 20 or 30 minutes before I killed it. | |
| 13:53:17 | finucannot | Tried with -vvv ? | |
| 13:53:23 | finucannot | I assume that will work | |
| 13:53:30 | fried_rice | was about to ask if there was some way to verbosify it. | |
| 13:53:32 | fried_rice | trying... | |
| 13:55:34 | fried_rice | Well, that's getting me more output, anyway. | |
| 13:55:39 | fried_rice | wtf is snowballstemmer?? | |
| 13:58:31 | fried_rice | finucannot: It looks like it's getting stuck really early, here: http://paste.openstack.org/show/719628/ | |
| 13:58:48 | finucannot | What version of pip are you on? | |
| 13:59:27 | fried_rice | finucannot: It looks like it's installing 10.0.1 into the venv | |
| 13:59:33 | melwitt | frickler: ah, sorry :( | |
| 13:59:43 | fried_rice | finucannot: under py27 | |
| 14:00:28 | finucannot | Yeah, that should be fine. I'm guessing this is a manifestation of some pypi.python.org -> pypi.org switchover bug | |
| 14:00:37 | finucannot | Weird that you're seeing it and no one else is though | |
| 14:00:52 | finucannot | Maybe just give up for today and try again Monday? :) | |
| 14:02:21 | fried_rice | finucannot: Well, maybe everybody else already has their venv created for docs. | |
| 14:02:43 | fried_rice | finucannot: I'm also still on trusty, don't know if that makes a difference. | |
| 14:03:10 | melwitt | frickler: and that's a good point about stable, I'll test it out | |
| 14:03:31 | finucannot | It shouldn't. That's why I asked what version of pip you were using but if it's installing everything from scratch, it can't be that | |
| 14:05:59 | finucannot | fried_rice: Yeah, works fine with '--recreate' on my end | |
| 14:06:23 | fried_rice | finucannot: Weird. Well, thanks for checking. | |
| 14:22:02 | melwitt | bauzas, superdan: could one of you please review this pcid cpu feature flags backport? need it to release ocata soon https://review.openstack.org/#/c/561168 | |
| 14:22:33 | superdan | yar | |
| 14:23:12 | melwitt | thx | |
| 14:23:37 | melwitt | also need this report allocated size of preallocated file based disks https://review.openstack.org/#/c/562078 pike change merged recently | |
| 14:23:47 | superdan | melwitt: on that one, he added _LW because backport but didn't line up the string | |
| 14:24:08 | superdan | melwitt: mriedem pointed it out kinda, but not sure if he was saying yeah this is fine or not | |
| 14:24:19 | superdan | melwitt: I got the pike one yesterday and was waiting for it to land | |
| 14:27:12 | melwitt | hm, yeah. "yar", that's usually a good thing | |
| 14:27:41 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP: placement: Granular GET /allocation_candidates https://review.openstack.org/517757 | |
| 14:30:03 | openstackgerrit | Merged openstack/nova master: Fix link in placement contributor doc https://review.openstack.org/563086 | |
| 14:30:20 | melwitt | superdan: yeah, not sure if he noticed that. we can ask him | |
| 14:31:23 | leakypipes | sahid: I'm really confused by the two patches that set driver name to None for vhost-user VIFs. Is there a bug that those patches are fixing? Also, I'm confused why there are two small patches where the second patch basically undoes part of the previous patch. | |
| 14:32:41 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: libvirt: Bump MIN_{LIBVIRT,QEMU}_VERSION for "Rocky" https://review.openstack.org/558783 | |
| 14:32:42 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: Drop MIN_LIBVIRT_SET_ADMIN_PASSWD https://review.openstack.org/562312 | |
| 14:33:48 | superdan | leakypipes: those are supposed to be hygeine patches for things were setting that are basically ignored right now by libvirt, but were creating confusing things underneath | |
| 14:34:07 | superdan | leakypipes: I dunno about the undoing-in-last-patch bit, but I asked those to be broken out because they were even more confusing being in the actual implementation patch | |
| 14:34:31 | superdan | leakypipes: we were setting driver=vhost even when type=vhostuser or something like that, and libvirt was just ignoring us | |
| 14:34:59 | leakypipes | superdan: ok. yes, I get that (as noted in my review comment). can those bottom two patches just be combined? | |
| 14:35:18 | superdan | leakypipes: I dunno, there was only one last I looked | |
| 14:35:36 | superdan | leakypipes: I want to see the vhostuser neutron job run against this if at all possible first though | |
| 14:35:51 | leakypipes | superdan: yes, saw that comment on the last patch in the series. | |
| 14:36:36 | superdan | leakypipes: ack cool | |
| 14:36:47 | superdan | I dunno why the bottom one was split, I haven't gone back to look at that yet today | |
| 14:36:49 | melwitt | lyaaaaaaaaarwood: want to hit this again? pike change merged https://review.openstack.org/#/c/560167 | |
| 14:37:09 | melwitt | and this https://review.openstack.org/#/c/560162 | |
| 14:39:26 | lyaaaaaaaaarwood | melwitt: ack | |
| 14:41:26 | kashyap | melwitt: Hey there, post milestone work busyness, also want to have a gander at this: https://review.openstack.org/#/c/558783/ | |
| 14:41:31 | kashyap | melwitt: You acked the other version bump thing. | |
| 14:41:47 | kashyap | (MattR already gave some feedback, I addressed it. Will let Zuul do its thing first) | |
| 14:41:57 | melwitt | superdan: here's another one for ocata https://review.openstack.org/#/c/545087 if you could pls take a look | |
| 14:43:20 | melwitt | kashyap: k, will put it on my list. still working on getting a couple of stable releases in line | |
| 14:43:43 | kashyap | melwitt: Sure, I just noticed I missed Matt's one more remark, I'll address that too | |
| 14:43:46 | kashyap | So disregard it for now | |
| 14:43:52 | melwitt | k | |
| 14:44:14 | sahid | leakypipes: all of this part is really confusing, some parts are tested some other not and there are conditions where we set the driver name and some where we don't set it... | |
| 14:45:03 | sahid | the two bottom patches are splitted because i wanted to make easy to read thing, in the first i'm enforcing the config to avoid setting driver name for vhostuser interface | |
| 14:45:23 | sahid | in the second i'm remove every pces where we set it wrongly | |
| 14:45:54 | sahid | superdan, leakypipes: about the test CI ... it makes sense I will work on it but I think it will be not possible to have it working | |
| 14:46:14 | sahid | we need version of QEMU/libvirt pretty recent and xenial provides old versions | |
| 14:46:21 | leakypipes | sahid: any chance of merging the bottom two patches? (sorry, and thanks for your patience) | |
| 14:46:24 | sahid | so at least we would see the logs | |
| 14:46:27 | superdan | sahid: I checked that job and it has libvirt 4.0 | |
| 14:46:32 | sahid | leakypipes: sure i will | |
| 14:46:37 | leakypipes | sahid: ty sir | |
| 14:47:22 | sahid | libvirt 4.0? i did not that version was released actually :) | |
| 14:48:11 | sahid | so i guess we are ok, did you get chance to look at the QEMU version? | |
| 14:49:25 | sahid | superdan: where did you find the info? | |
| 14:52:59 | superdan | sahid: in the devstack log, maybe I was looking at the wrong thing, but I'll show you what I found, just a sec | |
| 14:53:23 | superdan | sahid: 2018-04-18 23:19:55.218 | Get:3 http://ubuntu-cloud.archive.canonical.com/ubuntu xenial-updates/queens/main amd64 libvirt0 amd64 4.0.0-1ubuntu7~cloud0 [1,245 kB] | |
| 14:53:29 | superdan | from http://52.27.155.124/portland/2018-04-17/559414/4/check/tempest-dsvm-ovsdpdk-nfv-networking-xenial/8c6c0a1/logs/devstacklog.txt.gz | |
| 14:54:44 | superdan | 2018-04-18 23:26:02.509 | Get:7 http://ubuntu-cloud.archive.canonical.com/ubuntu xenial-updates/queens/main amd64 qemu amd64 1:2.11+dfsg-1ubuntu5~cloud0 [210 kB] | |
| 14:54:48 | superdan | qemu 2.11 reportedly | |
| 14:55:29 | superdan | sahid: looks to me like 4.0 was out in january of this year | |
| 14:57:55 | sahid | superdan: ok that looks perfect | |
| 14:59:13 | sahid | leakypipes: i just noticed you comment | |
| 14:59:19 | sahid | actually the libvirt doc is wrong | |
| 14:59:36 | sahid | the attribute is not ignored for vhostuser interface | |
| 14:59:57 | sahid | well, it is ignored is you set qemu or vhost backend (which doe not make any sense) | |
| 15:00:21 | sahid | setting vhost-user (which is what is passed to QEMU via -netdev vhost-user) does not work | |
| 15:00:31 | sahid | like foobar or everything | |
| 15:01:56 | sahid | in our current code, it happen that we pass 'vhost' that when we enable virtio multiqueue but it's wrong to set vhost | |
| 15:02:04 | sahid | so we have decided to clean all of that | |
| 15:02:26 | sahid | and never set a specific bakend for vhostuser interface | |
| 15:05:03 | sahid | hum.. it's all confusing and my phrasing make things ever more confusing, sorry about that. | |
| 15:12:58 | tomtom001 | Hello, I have an instance I'm resizing that keeps getting migrated to another CPU host when I do. Is that normal behaviour? Is it supposed to be doing that? | |
| 15:18:09 | melwitt | tomtom001: yes, resize == cold migrate | |