| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-08-22 | |||
| 13:33:38 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tests: Add '_setup_compute_services' helper https://review.opendev.org/663102 | |
| 13:33:57 | belmoreira | the was I'm running novnc is global, so is not connecting to the db to check the token | |
| 13:34:14 | belmoreira | s/was/way | |
| 13:34:27 | adrianc | stephenfin, sean-k-mooney : do we have a verdict on: https://review.opendev.org/#/c/675776/ ? | |
| 13:35:20 | sean-k-mooney | we could revert it and not that this is need due to a bug in the sriov nic agent | |
| 13:36:38 | stephenfin | belmoreira: melwitt will hopefully correct me if I say anything wrong, but '[workarounds] enable_consoleauth' enables a fallback path to help with migration to the new DB-based system | |
| 13:36:43 | sean-k-mooney | there is nolonger a need to set the mac for libvirt but it valid to still set it untill the sriov nic agent is updated. it is a bug that the agent depens on the mack however | |
| 13:37:19 | stephenfin | Emphasis on fallback. We'll use the DB tokens first and foremost and only fallback if no tokens are found | |
| 13:37:34 | adrianc | "a historic design decision" :) | |
| 13:37:37 | belmoreira | stephenfin I'm trying to use that without success. | |
| 13:38:31 | sean-k-mooney | adrianc: the historic design descis being the only thing that was allowed to be depeded on is the pci address :) | |
| 13:38:41 | sean-k-mooney | at somepoint i guess neutron forgot that | |
| 13:39:31 | sean-k-mooney | the sriov nic agent is in principal ment to be able to work with non libvirt virt drivers too | |
| 13:39:33 | belmoreira | stephenfin token validation fails | |
| 13:40:08 | sean-k-mooney | like xen or powervm although i dont know if it is used with either | |
| 13:40:26 | yonglihe | sean-k-mooney: thanks given your valuable +1 for https://review.opendev.org/#/c/621476/ | |
| 13:40:50 | stephenfin | belmoreira: Odd. I'm looking at commit 969239029d4a13956747e6e0b850d6c6ab4035f0 rn to see what could possibly be going wrong | |
| 13:41:36 | stephenfin | belmoreira: Dumb question, but you have this option set on the conf files used by the 'nova-novncproxy' service, right? | |
| 13:41:51 | belmoreira | stephenfin yes | |
| 13:41:58 | stephenfin | Cool. Just checking :) | |
| 13:42:01 | sean-k-mooney | stephenfin: feel like sending https://review.opendev.org/#/c/621476/ on its way and closing out that blueprint | |
| 13:42:48 | stephenfin | I'd been holding off on that because I still wasn't sure how I feel about that whole thing, heh | |
| 13:43:02 | stephenfin | But yeah. Let me finish looking into this consoleauth thing first though | |
| 13:43:19 | sean-k-mooney | ok if you have time to add it to your review queue then its an easy win assumeing you decide how you feel about it | |
| 13:43:33 | adrianc | sean-k-mooney: ill open a bug in neutron, once the revert is in ill abandon https://review.opendev.org/#/c/676713 | |
| 13:44:31 | sean-k-mooney | adrianc: why abandon that? it would be good to open the bug first and update the comment to reference the new bug | |
| 13:44:43 | sean-k-mooney | oh because it still depens on the mac | |
| 13:44:56 | sean-k-mooney | jsut looks at a different source | |
| 13:45:07 | adrianc | yah, we talked that changes will be required from server side | |
| 13:45:16 | adrianc | i.e the mechanism driver | |
| 13:46:19 | stephenfin | belmoreira: Have you tracebook to look at? There are a couple of InvalidToken exceptions that could be raised in that code | |
| 13:46:23 | stephenfin | *trackback | |
| 13:46:25 | stephenfin | *traceback | |
| 13:46:28 | stephenfin | third time lucky | |
| 13:48:01 | openstackgerrit | Alexandre arents proposed openstack/nova master: Abort live-migration during instance_init https://review.opendev.org/678016 | |
| 13:48:51 | stephenfin | I can't see anything obvious there, though I was wrong with my statement about it being a fallback. If that's toggled, we check the consoleauth service first, not after | |
| 13:51:14 | belmoreira | stephenfin: from consoleauth: Checking Token: ***, False | |
| 13:52:22 | shilpasd | gibi: Hi, i have pushed revised patches (waiting for zuul) but i have lost your +2s, requesting you to review/vote further, thanks | |
| 13:53:40 | shilpasd | Addressed your review comments and uploaded patches for isolated aggregates patches. | |
| 13:54:02 | belmoreira | stephenfin: from nova-novncproxy: https://www.codepile.net/pile/O1P8ee1q | |
| 13:54:55 | stephenfin | belmoreira: Should I see something there? | |
| 13:55:55 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Update docstring of 'revert_resize' function https://review.opendev.org/665976 | |
| 13:56:32 | belmoreira | stephenfin: the way that is deployed is global. nova-novncproxy and consoleauth are not deployed per cell | |
| 13:57:54 | stephenfin | Oh, that could be problematic | |
| 13:58:04 | belmoreira | stephenfin: that is the trace that I get. the "code 404, message File not found" is new to me | |
| 14:00:03 | efried | nova meeting now | |
| 14:00:08 | Sundar | dansmith, sean-k-mooney, mriedem, gibi: Can you please review https://review.opendev.org/#/q/status:open+project:openstack/nova+bp/nova-cyborg-interaction ? I have added the notification and addressed the main comments. Will add more UT, tempest CI framework is making progress. | |
| 14:00:43 | Sundar | Well, all are welcome to review. | |
| 14:00:45 | stephenfin | belmoreira: I think you're going to need to open a bug for this. I don't know if this has been discussed as an issue before (I don't recall it having been so) | |
| 14:01:10 | efried | sean-k-mooney: We were going to wait until the name was official to start creating the artifacts in lp and nova-specs. If you have a spec you want to open right now, you can write it against backlog/ and then move it. | |
| 14:01:26 | sean-k-mooney | Sundar what os are you deveoping on? i can try and deploy the series in a vm | |
| 14:01:49 | efried | alex_xu, sean-k-mooney, aspiers: request filter stuff after the meeting? | |
| 14:01:59 | alex_xu | efried: yea | |
| 14:02:03 | sean-k-mooney | it will be next week before i can spend a lot of time on ith however | |
| 14:02:05 | belmoreira | stephenfin ok thanks | |
| 14:02:18 | sean-k-mooney | efried: ya i can do that. its going to be really short anyway | |
| 14:03:21 | sean-k-mooney | efried: stephenfin and i have an internal team meeting after the nova meeting but we will be still on irc | |
| 14:03:25 | Sundar | sean-k-mooney: It is Centos 7.6, kernel 3.10.0-957 | |
| 14:03:34 | Sundar | Thanks | |
| 14:03:54 | sean-k-mooney | Sundar: im assuming you testing with python 2 still? or python 3 | |
| 14:04:04 | Sundar | Yes, python 2 | |
| 14:04:35 | sean-k-mooney | ok i have been having issue with centos and python2 lately but ill spin up a vm and try and deploy it | |
| 14:04:50 | belmoreira | stephenfin is there any recommendation how to setup nonvc with multiple cells? | |
| 14:04:55 | aspiers | efried: sure | |
| 14:05:07 | sean-k-mooney | the fake driver merged in cyborg right so i should be able to test with that | |
| 14:05:07 | dansmith | belmoreira: I think you really want to talk to melwitt | |
| 14:05:08 | belmoreira | stephenfin I'm thinking how this should work now. We will need to have nova-novncproxy available per cell. This means that the vnc url that I will get will include the cell controller. Is this correct? | |
| 14:08:53 | Sundar | sean-k-mooney: Yes, the fake driver has merged. Please deploy https://review.opendev.org/#/q/project:openstack/cyborg+branch:master+topic:nova-integ+status:open . The part where Cyborg reports to Placement, https://review.opendev.org/#/c/659233/, is also WIP. But the combination works today with some massaging. We will sort out the wrinkles before | |
| 14:08:54 | Sundar | merging these patches. | |
| 14:08:57 | stephenfin | belmoreira: I really don't know, I'm afraid :( melwitt is your person | |
| 14:09:12 | belmoreira | dansmith stephenfin thanks, will try to talk with melwitt | |
| 14:09:13 | stephenfin | I know you'll need a service per cell, yes, but how that will look I'm not sure | |
| 14:09:36 | stephenfin | https://github.com/openstack/devstack/blob/master/lib/nova#L947-L973 | |
| 14:17:02 | openstackgerrit | Adrian Chiris proposed openstack/nova master: Revert "vif: Resolve a TODO and update another" https://review.opendev.org/675776 | |
| 14:30:46 | mriedem | dtantsur: TheJulia: did you find out if https://review.opendev.org/#/q/topic:bug/1839560+branch:stable/stein fixed your grenade issue with rebalancing? | |
| 14:32:14 | TheJulia | mriedem: upon issuing a change on master branch, zuul immediately comes back and says there is a branch merge failure with dependencies of a dependent patch on stable/stein | |
| 14:32:28 | TheJulia | It used to work, but... I guess not anymore. | |
| 14:33:43 | aspiers | sean-k-mooney: recheck is apparently not working for https://review.opendev.org/#/c/666915/ | |
| 14:34:48 | stephenfin | yonglihe, sean-k-mooney: Think there's a bug with that NUMA API change | |
| 14:34:52 | stephenfin | comments left | |
| 14:35:00 | sean-k-mooney | it hit a post failure again or there was a zuul restart and it got droped im guess | |
| 14:35:16 | stephenfin | tl;dr: I'd kill the summary view entirely. I don't think it's possible for it to be reliable | |
| 14:35:19 | sean-k-mooney | stephenfin: ya just read good cache on the mutable data in kwargs | |
| 14:35:28 | sean-k-mooney | stephenfin: i though we had a hacking check for that | |
| 14:35:39 | yonglihe | stephenfin: post me a comment, i gonna fix it soon. and you got new update next day. thanks. | |
| 14:35:57 | stephenfin | yonglihe: Cool. Thanks :) | |
| 14:39:12 | yonglihe | stephenfin: for most important of -1, it's we should drop instance.vcpu_model, right? | |
| 14:39:32 | stephenfin | yonglihe: In a meeting. Give me 10 | |
| 14:39:45 | yonglihe | sure. | |
| 14:46:28 | stephenfin | yonglihe: to be honest, I'd personally like to drop the 'cpu_topology' element [1] from the response [1] https://review.opendev.org/#/c/621476/57/doc/api_samples/os-server-topology/v2.77/servers-topology-resp-user.json@2 | |
| 14:46:44 | stephenfin | so we'd just include the 'nodes' element | |
| 14:47:20 | stephenfin | yonglihe: The reason being asymmetric topologies, which we support | |
| 14:47:47 | stephenfin | I can do this: | |
| 14:48:03 | yonglihe | Ok, got your concern. That's seems very right thing to do. | |
| 14:48:14 | stephenfin | openstack flavor create --vcpus 3 --ram 768 --disk 20 test.asymmetric | |
| 14:49:28 | stephenfin | openstack flavor set test.asymmetric --property 'hw:numa_nodes=2' --property 'hw:numa_cpus.0=0' --property 'hw:numa_cpus.0=1,2' --property 'hw:numa_mem.0=256' --property 'hw:numa_mem.1=512' | |
| 14:49:28 | yonglihe | I know that stuff, but it slipped away while wrote spec -:). | |
| 14:49:34 | stephenfin | Okay, sweet :) | |
| 14:50:11 | stephenfin | Also, I wonder if we should rename 'vcpu_set' to 'cpus' ? | |
| 14:50:18 | yonglihe | I bet alex_xu also buy in this solution. | |