| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-24 | |||
| 13:02:02 | kashyap | macsz: Maybe email him or, if it's a public Nova issue, write to upstream openstack-dev list, and CC him | |
| 13:02:12 | johnthetubaguy | andreykurilin: I am running to lunch, but do we know which patch caused the problem? | |
| 13:02:16 | macsz | kashyap: thanks | |
| 13:02:37 | andreykurilin | johnthetubaguy: bon appetit! | |
| 13:03:49 | sfinucan | johnthetubaguy: An easy, post-merge conflict +2/+W for you (for after lunch :)) https://review.openstack.org/408193 | |
| 13:03:52 | andreykurilin | johnthetubaguy: for python-api - my patch which removes custome implementation of HTTPClient made it. there were a bunch of patches with + from keystone cores, but we missed to transmit several variables to keystone Token class | |
| 13:04:18 | openstackgerrit | Steven Webster proposed openstack/nova: Release PCI devices on drop_move_claim() https://review.openstack.org/370374 | |
| 13:04:57 | andreykurilin | johnthetubaguy: as for CLI, imo, https://github.com/openstack/python-novaclient/commit/1f11840dd84f3570330d1fcd53d1e8eea5ff7922 broke it. more than a year ago.. | |
| 13:05:22 | takashin | bauzas: Are you around? | |
| 13:06:45 | mdbooth | Do we currently recalculate numa topology on live migration? | |
| 13:07:01 | andreykurilin | johnthetubaguy: in case of python-API: previously, Token auth method worked only via our custom HTTPClient implementation. It means that only Keystone V2 was supported. Switch to SessionClient for 100% cases was done here - https://review.openstack.org/#/c/304035/ | |
| 13:08:13 | sfinucan | mdbooth: Narp | |
| 13:08:24 | sfinucan | mdbooth: Isn't that the patch you were reviewing last night? | |
| 13:08:28 | mdbooth | sfinucan: Ok, cool. | |
| 13:08:40 | mdbooth | sfinucan: I'm reviewing it now, but I was wondering if it clashed with anything :) | |
| 13:09:27 | sfinucan | Aha, no - you're good. None of that stuff is recalculated at present, leading to a host of bugs around CPU pinning, hugepages and anything else NUMA-affinitized | |
| 13:09:50 | mdbooth | sfinucan: Have you reviewed this, btw? | |
| 13:10:47 | sfinucan | Aye, multiple times. I never made it to the end though before it got respun :) | |
| 13:11:50 | mdbooth | I'm assuming that means that, priority list or not, it's not making it into Ocata. | |
| 13:12:00 | mdbooth | It's 'one of those' | |
| 13:13:05 | mdbooth | sfinucan: Incidentally, how do guests feel about having their instance topologies changed? | |
| 13:13:24 | mdbooth | sahid: Do you know the answer to ^^^ ? | |
| 13:13:51 | sfinucan | mdbooth: idk - it really needs to get in but we keep finding corner cases (race conditions, etc.) | |
| 13:14:06 | sfinucan | ...and it's so damn big that reviewing it turns into a chore each time | |
| 13:14:10 | mdbooth | Where have the races been? | |
| 13:14:19 | mdbooth | There's one in there now, btw. I already commented on that. | |
| 13:14:34 | mdbooth | It's trivial to fix, though, just a missing sync context manager | |
| 13:14:48 | sfinucan | cfriesen pointed out a few some time back. I don't know if they were addressed | |
| 13:15:48 | sfinucan | mdbooth: As for the instance topology, I don't think the actual topologies change | |
| 13:15:59 | sfinucan | i.e. how many sockets/cores/threads everything has | |
| 13:16:09 | sfinucan | rather, how these are aligned under the hood | |
| 13:16:15 | sfinucan | on the host, that is | |
| 13:19:28 | sfinucan | I could be wrong there though. The scheduler is still mostly black magic to me :( | |
| 13:20:30 | sfinucan | mdbooth: numa_fit_instance_to_host takes an already defined instance_topology and shouldn't change anything about that (besides pinning information, which is the bug I'm sidestepping here https://review.openstack.org/#/c/419479/) | |
| 13:20:50 | mdbooth | sfinucan: Cool. | |
| 13:23:14 | openstackgerrit | Stephen Finucane proposed openstack/nova: pci: Validate behavior of empty devname https://review.openstack.org/396130 | |
| 13:32:32 | openstackgerrit | Radomir Dopieralski proposed openstack/python-novaclient: Pass relevant parameters to Token based authentication https://review.openstack.org/419441 | |
| 13:44:01 | openstackgerrit | Gábor Antal proposed openstack/nova: Added instance.reboot.error to the legacy notifications https://review.openstack.org/411816 | |
| 13:50:07 | bauzas | johnthetubaguy: not sure I see which points in particular you want to change for https://review.openstack.org/#/c/417961/20 ? | |
| 13:50:32 | bauzas | johnthetubaguy: just a naming thing, or more about how to not return a NoValidHosts in case we don't have the placement service ready yet ? | |
| 13:50:50 | johnthetubaguy | bauzas: the naming thing was my main worry | |
| 13:50:57 | bauzas | the latter needs discussion because I think the general agreement was against that | |
| 13:51:05 | bauzas | but that's tied to the grenade issue | |
| 13:51:31 | bauzas | maybe an hangout could help with folks around | |
| 13:51:52 | bauzas | but we would be missing jaypipes and matt | |
| 13:51:57 | bauzas | :( | |
| 13:52:04 | bauzas | in the meantime, I can do the naming thing | |
| 13:52:33 | cdent | bauzas, johnthetubaguy : I'm not yet caught up on that review, but will be within about two hours. Will try to add my perspective, if it differs from anything already there. | |
| 13:52:51 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Pass relevant parameters to Token based authentication https://review.openstack.org/419441 | |
| 13:53:05 | openstackgerrit | Gábor Antal proposed openstack/nova: Transform instance.reboot.error notification https://review.openstack.org/411791 | |
| 13:53:10 | bauzas | cdent: refrain your pencil, I have a patch to do for modifying the interface name | |
| 13:53:26 | cdent | ✔ | |
| 13:53:34 | openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: [CLI] Fix token auth type https://review.openstack.org/424256 | |
| 13:54:22 | andreykurilin | bauzas: hi! please look at https://review.openstack.org/#/c/419441 it is critical fix for several projects | |
| 13:59:43 | openstackgerrit | Gábor Antal proposed openstack/nova: Transform instance.reboot.error notification https://review.openstack.org/411791 | |
| 13:59:43 | takashin | bauzas: Could you review https://review.openstack.org/#/c/408955/ again and check my comments? | |
| 13:59:53 | tdurakov | paul-carlton2 eliqiao shaohe_feng pkoniszewski yuntongjin mdbooth kashyap kmartin claudiub tobasco tdurakov andreas_s abhishekk mrhillsman raj_singh wznoinsk siva_krishnan hi folks, live migration subteam meeting in a minute | |
| 14:01:17 | takashin | bauzas: Are you around? | |
| 14:01:34 | johnthetubaguy | bauzas: I think I don't really understand fully what the current proposal for update is / means, so its hard to comment on it. | |
| 14:02:14 | bauzas | johnthetubaguy: that's why I think an hangout could be nice | |
| 14:02:28 | takashin | bauzas: Could you review https://review.openstack.org/#/c/408955/ again and check my comments? | |
| 14:02:49 | bauzas | johnthetubaguy: so you could give your insight once knowing all | |
| 14:03:05 | johnthetubaguy | bauzas: yeah, that could work | |
| 14:03:06 | bauzas | johnthetubaguy: and honestly, I tend to listen to your voice for operating insights :) | |
| 14:03:24 | bauzas | takashin: yup, in my queue, I haven't forgotten | |
| 14:03:34 | bauzas | just feeling the FF breeze chilling me | |
| 14:03:50 | takashin | bauzas: Thank you. | |
| 14:04:24 | openstackgerrit | Radomir Dopieralski proposed openstack/python-novaclient: ListExtResource given in place of ListExtManager https://review.openstack.org/424570 | |
| 14:04:58 | bauzas | unrelated, having the UTs running with osprofiler anytime seems sooo wrong to me | |
| 14:05:05 | bauzas | DinaBelova: around ? | |
| 14:05:26 | bauzas | DinaBelova: why are we needing to verify osprofiler anytime we run U | |
| 14:05:28 | bauzas | UTs? | |
| 14:05:40 | bauzas | can't it be part of the regular unittest job? | |
| 14:06:06 | bauzas | as a developer, I don't really need to verify that anytime I'm modifying an internal change | |
| 14:06:15 | bauzas | that's... silly | |
| 14:06:31 | bauzas | but I truly understand it could be part of the UT coverage | |
| 14:06:41 | bauzas | or a specific job | |
| 14:11:28 | jsheeren | hi all, the default flavors in mitaka and earlier had a naming like m1.small etc, what is the origin of this? to be similar to amazon? | |
| 14:12:54 | andreykurilin | ACHTUNG! Nova's intergration with Horizon, Trove, Murano-dashboard, Cinder(?!) is broken - https://review.openstack.org/#/c/419441 | |
| 14:13:18 | openstackgerrit | Sylvain Bauza proposed openstack/nova: WIP: Scheduler calling the Placement API https://review.openstack.org/417961 | |
| 14:13:28 | andreykurilin | jaypipes ^ | |
| 14:14:17 | bauzas | andreykurilin: on it | |
| 14:14:27 | openstackgerrit | Gábor Antal proposed openstack/nova: Transform instance.volume_attach notification https://review.openstack.org/401992 | |
| 14:14:33 | andreykurilin | bauzas: thanks :) | |
| 14:14:40 | bauzas | andreykurilin: AFAIK, it's impacting TripleO as well, so I'd recommend to tag the bug as critical | |
| 14:15:06 | andreykurilin | bauzas: done | |
| 14:17:39 | bauzas | andreykurilin: diana_clarke: there is a follow-up for fixing CLI, correct? | |
| 14:17:47 | diana_clarke | bauzas: yup | |
| 14:17:52 | andreykurilin | bauzas: https://review.openstack.org/#/c/424256 | |
| 14:17:58 | bauzas | okay, let's merge the first one and see the CLI then | |
| 14:18:06 | andreykurilin | bauzas: it is broken for more than a year :( | |
| 14:18:17 | bauzas | diana_clarke: could you please mark your change as Related-bug ? | |
| 14:18:47 | diana_clarke | ^ andreykurilin | |
| 14:19:02 | diana_clarke | (I don't have a change to mark as related) | |
| 14:19:21 | andreykurilin | bauzas: I think we do not have a right bug-report for CLI, but I can create it for proper history | |
| 14:19:39 | bauzas | andreykurilin: oh wait, the bug is 6 month old ? | |
| 14:20:06 | bauzas | if so, that's definitely not a critical :p | |
| 14:20:10 | andreykurilin | lol | |
| 14:20:12 | andreykurilin | bauzas: token auth via python API is broken since 7.0.0 | |
| 14:20:20 | andreykurilin | it is just several weeks | |