| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-23 | |||
| 13:33:45 | bauzas | cdent: you'll see now the conditional branch for checking the filters before calling the placement service | |
| 13:34:47 | bhagyashris | cdent: I have few questions in my mind regarding Scheduler calling the Placement API [1] https://review.openstack.org/#/c/417961. 1. As it updates allocation information at the placement db so is there a need of compute node table for resource tracking? | |
| 13:37:07 | cdent | bhagyashris: the resource tracker and the nova-scheduler will carry on needing to know about compute nodes, yes. Is that what you're asking? The placement services itself will not need to know about the compute node table, just the uuids of those compute nodes, represented as resource providers in the placement api | |
| 13:37:48 | openstackgerrit | Mikhail Feoktistov proposed openstack/nova: libvirt: ephemeral disk support for virtuozzo containers https://review.openstack.org/312488 | |
| 13:42:29 | openstackgerrit | Claudiu Belu proposed openstack/nova: hyper-v: Do not allow instances with pinned CPUs to spawn https://review.openstack.org/424116 | |
| 13:47:18 | sfinucan | lyarwood: Done and done https://review.openstack.org/#/c/418371/1 | |
| 13:47:45 | sfinucan | johnthetubaguy: ^ that looks like a straightforward +W - could you take a glance if you have a moment? | |
| 13:49:13 | lyarwood | sfinucan: cheers | |
| 13:50:22 | bhagyashris | cdent: I know , i am asking lot of questions :) I just want to know if we are using the placement apis then will compute node table be needed in that case? If it will used then in which case it will be used and for what? | |
| 13:51:29 | diga | jaypipes: Hi | |
| 13:51:31 | johnthetubaguy | sfinucan: ack, thanks | |
| 13:51:31 | cdent | bhagyashris: it will carry on being used for keeping track of the compute nodes because nova still needs to know about them | |
| 13:51:51 | cdent | diga: jaypipes is on PTO, maybe somebody else can help? | |
| 13:52:04 | diga | cdent: Hi, sure | |
| 13:52:20 | diga | cdent: https://review.openstack.org/#/c/423872/ can you review this spec - placement notification | |
| 13:53:09 | cdent | yeah, sure, will stick that on my list for today | |
| 13:54:22 | diga | cdent: :) thank you! | |
| 13:54:31 | diga | cdent: we have meeting now right ? | |
| 13:54:36 | cdent | yeah, in five minutes | |
| 13:55:00 | diga | okay | |
| 13:55:04 | edleafe | diga: in #openstack-meeting-alt | |
| 13:55:21 | diga | edleafe: Sure, will join | |
| 13:56:07 | andreykurilin | amrith: hi | |
| 13:56:23 | amrith | hi andreykurilin | |
| 13:56:33 | amrith | did you see the email I just sent you | |
| 13:56:47 | andreykurilin | yes | |
| 13:56:50 | johnthetubaguy | bauzas: you say only ironic needs the host field, isn't that what we use for the RPC call? | |
| 13:57:05 | amrith | the issue I'm facing is that if I use the v6.0.0 client things work fine | |
| 13:57:15 | amrith | if I switch to the v7.0.0 client, it bombs out | |
| 13:57:21 | bauzas | johnthetubaguy: I meant the problem is that host != node for Ironic | |
| 13:57:23 | johnthetubaguy | bauzas: or is that the node vs host thing | |
| 13:57:29 | amrith | am not sure why that would be the case | |
| 13:57:29 | bauzas | yup that | |
| 13:57:30 | johnthetubaguy | bauzas: ah, gotcha | |
| 13:57:36 | bauzas | maybe my comment is unclear, sorry if so | |
| 13:57:38 | andreykurilin | amrith: I have just several minutes before meeting and will able to help you in ~30 minutes | |
| 13:57:48 | bauzas | but I can't say that I could just use the RP name | |
| 13:58:01 | bauzas | given it's the nodename which can be different for ironic | |
| 13:58:05 | amrith | andreykurilin, I'll be here. | |
| 13:58:13 | andreykurilin | amrith: but now I have one question: What is user.tenant https://github.com/openstack/trove/blob/master/trove/tests/util/__init__.py#L170 here? Is it tenant name or id? | |
| 13:58:23 | bauzas | while the other drivers (except the Fake driver, but that's fixed in a bottom patch) are using host == node | |
| 13:58:29 | bauzas | johnthetubaguy: ^ | |
| 13:58:31 | amrith | looking | |
| 13:59:56 | johnthetubaguy | bauzas: the custom resource class stuff should fix that, I am assuming | |
| 14:00:13 | bauzas | johnthetubaguy: yeah, but it could be merged by Pike | |
| 14:00:23 | bauzas | johnthetubaguy: hence me using the new object method I created | |
| 14:00:44 | bauzas | we could just cut that once the above BP you mentioned is fully merged | |
| 14:01:00 | bauzas | but AFAICS, that's still not done yet hence my big fat comment | |
| 14:01:14 | johnthetubaguy | bauzas: yeah, no biggie, just getting my head straight on that | |
| 14:01:40 | bauzas | np | |
| 14:01:55 | bauzas | johnthetubaguy: I have a bigger concern that I wrote in a ML thread this week-end | |
| 14:02:03 | edleafe | bauzas: scheduler meeting? | |
| 14:02:11 | bauzas | edleafe: I can do, for 20 mins | |
| 14:02:21 | edleafe | cool | |
| 14:02:23 | amrith | andreykurilin, one second, you are looking at the tests. I wasn't running test code | |
| 14:02:32 | bauzas | edleafe: but I'm pretty down given all the work I'm doing atm | |
| 14:02:33 | andreykurilin | ok... | |
| 14:02:34 | amrith | I was running a straight trove command which tried to hit nova | |
| 14:02:42 | amrith | I will find out what that is | |
| 14:02:58 | andreykurilin | I just look at first failure at gates:) | |
| 14:03:01 | andreykurilin | *looked | |
| 14:03:12 | amrith | but for what I've been trying to do, the code is at https://github.com/openstack/trove/blob/master/trove/common/remote.py#L101 | |
| 14:03:17 | amrith | ah, sorry | |
| 14:03:27 | amrith | I am running trove locally to debug | |
| 14:03:38 | amrith | here, we're making a client with version 2.12 | |
| 14:03:58 | amrith | which comes from https://github.com/openstack/trove/blob/master/trove/common/cfg.py#L67 | |
| 14:04:05 | amrith | and we're passing in a bunch of parameters | |
| 14:04:23 | amrith | my specific issue (and like I said in the email, trove may be doing something wrong here) | |
| 14:04:33 | amrith | we're making the call like https://github.com/openstack/trove/blob/master/trove/common/remote.py#L101 | |
| 14:04:49 | amrith | and if I understand compatibility right, this should work unchanged with the v7.0.0 client from nova | |
| 14:04:51 | amrith | but it isn't | |
| 14:04:51 | andreykurilin | amrith: need to go to meeting, sorry. PS: it should help(with depends-on) ttps://review.openstack.org/412497 | |
| 14:05:09 | amrith | nope, doesn't help | |
| 14:05:15 | amrith | tried it | |
| 14:05:36 | amrith | something else is broken, not sure where | |
| 14:05:51 | amrith | the issue I'm facing is that nova returns a 403 (forbidden) on a simple thing like flavor list | |
| 14:05:54 | andreykurilin | amrith: I added depends-on there | |
| 14:06:17 | bhagyashris | cdent: Ok. Thank you. :) | |
| 14:06:21 | andreykurilin | amrith: did you check it with pulling not-merged changed from novcliane? | |
| 14:06:25 | amrith | oh, sorry | |
| 14:06:27 | amrith | different change | |
| 14:06:30 | amrith | let me look | |
| 14:06:43 | amrith | NO, I did not check with pulling not merged change in novaclient | |
| 14:06:48 | amrith | that may be it. let me check | |
| 14:06:53 | amrith | catch yo after your meeting | |
| 14:06:54 | amrith | thanks! | |
| 14:07:34 | alex_xu | johnthetubaguy: sdague, hope you can take a look at this patch https://review.openstack.org/#/c/423952/5, we break some old microversion in newton release. cdent have a good point on it. I'm not sure we can fix it directly, maybe another microversion? | |
| 14:08:20 | johnthetubaguy | alex_xu: yeah, sounds like it should be added back in a the next microversion | |
| 14:10:04 | alex_xu | johnthetubaguy: yea, we already bump few versions after break that. | |
| 14:14:13 | sdague | alex_xu: so, before fixing it, is there something we can learn about how to not let something like that fall through review again? | |
| 14:15:05 | sdague | no one seemed to catch that over a lot of review iterations (64 revisions on that patch, that bug showed up in revision 44) | |
| 14:15:13 | sdague | https://review.openstack.org/#/c/304510/44/nova/api/openstack/compute/block_device_mapping.py | |
| 14:15:43 | openstack | Launchpad bug 1658148 in OpenStack Dashboard (Horizon) "Horizon/Nova interactions broken " [Critical,New] - Assigned to Rob Cresswell (robcresswell) | |
| 14:15:43 | diana_clarke | amrith, andreykurilin: Horizon is reporting a similar issue with python-novaclient 7.0.0. https://bugs.launchpad.net/python-novaclient/+bug/1658148 | |
| 14:16:04 | alex_xu | sdague: i'm thinking is there any test can cover that. we probably only run the test on specific version, not in the future version. | |
| 14:17:28 | alex_xu | but even the test can run in the future version, but the bug only can be found when we bump a new version | |
| 14:18:56 | sdague | alex_xu: right, I was wonder if we could iterate through the schema and compare at various versions | |
| 14:19:13 | sdague | including a fake new version after the one being committed | |
| 14:19:23 | sdague | and if we see a blip and back, that's probably wrong | |
| 14:19:24 | andreykurilin | amrith: ok, I returned | |