Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-02-24
15:12:50 elod lyarwood: i can try to do that (i've never done before)
15:13:31 gibi stephenfin: I will start with those tomorrow (currently deep in a pike re-schedule bug downstream)
15:13:40 elod lyarwood: is that enough for the release or are there anything else that need to be done?
15:14:20 lyarwood elod: that's it AFAIK, efried_pto will need to bless it as PTL but that's about it
15:16:30 elod lyarwood: sounds good \o/ then i'll create the patch and add you then efried_pto as reviewer
15:18:00 lyarwood elod: awesome thanks
15:34:06 elod lyarwood: here it is: https://review.opendev.org/#/c/709554/ i think 18.2.4 will be fine, but please double check
15:53:22 bauzas efried_pto: stephenfin: thanks for reviewing https://review.opendev.org/#/c/589085/
15:53:50 bauzas FWIW, I haven't yet provided a ML thread about this because I also want to fix the resize revert
15:53:59 bauzas so, I'll provide two other changes
15:54:14 bauzas once it's done, I'll also provide a ML thread
17:27:03 openstackgerrit Merged openstack/nova master: Allow TLS ciphers/protocols to be configurable for console proxies https://review.opendev.org/679502
19:48:40 openstackgerrit melanie witt proposed openstack/nova stable/pike: Fix os-simple-tenant-usage result order https://review.opendev.org/709600
20:00:32 openstackgerrit melanie witt proposed openstack/nova stable/ocata: Fix os-simple-tenant-usage result order https://review.opendev.org/709603
#openstack-nova - 2020-02-25
00:49:49 sean-k-mooney dansmith: i think https://review.opendev.org/#/c/709641/2 will create a tempest full multi node job with a cyborg flavor. it will take a while too run but it should be close if it does not fully work so ill take a look at it agian tomorow
02:11:11 openstackgerrit wangjiajing proposed openstack/nova master: Perfect parameter of method. https://review.opendev.org/709652
03:18:09 openstackgerrit Merged openstack/nova master: Deprecate base rules in favor of new rules https://review.opendev.org/701624
05:28:38 openstackgerrit Arthur Dayne proposed openstack/nova master: libvirt:driver:Disallow AIO=native when 'O_DIRECT' is not available https://review.opendev.org/682772
05:29:03 openstackgerrit MIKI Nobuhiro proposed openstack/nova master: Enable instance to be booted with neutron SR-IOV ports https://review.opendev.org/709658
06:05:43 rm_work ileixe: hey, thanks for your review of my routed_segments filter
06:06:09 ileixe rm_work: :)
06:06:12 rm_work ileixe: do you think I should remove the doc/source/user/ change since it is intended for admin use only?
06:06:40 rm_work I was just following the previous set examples, i didn't see a case where a filter was not in both
06:07:37 rm_work and, I am not sure about how this filter would interact with Ironic, as we are not using it with Ironic (godaddy does not deploy Ironic, and Verizon Media has a TON of Ironic usage, but never needing this filter)
06:08:12 rm_work on the patch I mentioned: The main two use-cases are for the Octavia service to keep Loadbalancer VMs on the same segment for failover purposes, and for Live Migrate in routed-networks (I'll be pushing a followup patch for that soon). Users generally will not have a need to use this.
06:09:18 ileixe rm_work: um.. i think yes, if you thought it's only for the use-cases.
06:09:48 rm_work OK, then I will remove that user-facing doc piece
06:09:52 ileixe Do you plan to use the filter for extended usecases?
06:10:12 rm_work it could be used for any number of things, but as you point out, non-admin users can't see segment-ids
06:10:21 rm_work so it would be limited in usefulness
06:10:55 ileixe Yep then I think it's better to abstract the use for normal user
06:11:04 rm_work I could add a thing maybe to allow a user to pass in a port-id or subnet-id? and we could fetch the port/subnet and get the segment?
06:11:37 ileixe Do you see the previous work for routed network? (https://review.opendev.org/#/c/656885/)
06:11:39 rm_work is it possible to do this just the first time and cache that data in the request_spec?
06:11:40 ileixe Did you see
06:12:21 rm_work i looked briefly but it is a lot more complicated, and it seemed it was not moving forward really
06:13:02 rm_work it's been open for almost a year and now missed another spec freeze...
06:13:15 rm_work this filter seems like it should be a much simpler thing to merge
06:13:53 ileixe Honestly, I think maybe it's better to add the functionality like that (make resource provider and use request_spec from neutron) for normal user.
06:14:05 ileixe But for admin, I think simple filter is also userful
06:14:09 ileixe *useful
06:14:42 rm_work ok, so limit this filter for admin like it is now, and let people add more advanced user functionality to the patch you linked (or a followup)?
06:14:53 ileixe Yes that's what I thought
06:15:29 ileixe So I wonder how others operate now since it seems to be open question
06:15:37 ileixe Are you working with GoDaddy?
06:15:58 rm_work i was there for two years previously, but now I have been at Verizon Media for a year
06:16:09 rm_work coincidentally, both use routed networks everywhere :D
06:16:23 ileixe Yes, I assume both of them are using a lot
06:16:26 rm_work probably because both are very large deployments
06:16:35 rm_work I am also the PTL for Octavia
06:16:38 ileixe How verizon use routed network for end-user?
06:16:48 ileixe Do they maually pre-make port?
06:16:52 rm_work and have been pushing a fix on that project to enable use in routed-networks
06:16:53 ileixe Oh great
06:17:07 rm_work end-user is generally unaware
06:17:12 rm_work because it does not matter for them in daily use
06:17:17 rm_work as long as "everything works"
06:17:44 rm_work but for things like Octavia, or live-migrate (which are both service-side uses), the routed-network segment is important
06:17:56 rm_work so again, the user can be shielded from knowing anything at all about it
06:18:07 rm_work as long as the service handles it in the backend
06:18:28 ileixe Yes, I think it's more important to other services like Octavia
06:18:38 ileixe In fact, I investigate it since I have to use ironic too
06:18:58 rm_work it's possible that Ironic could be patched to make use of this filter too, I assume
06:19:00 ileixe There seems to be no way to choose segment
06:19:04 ileixe in ironic
06:19:05 ileixe yes
06:19:08 rm_work I just don't know much about it
06:19:17 rm_work so yes, if we could merge this filter, it would make things like that possible
06:21:34 ileixe Yes, hope some other guys push forward
06:25:37 rm_work updating the patch to make it clearly admin-only
06:25:44 rm_work thanks for your feedback!
06:26:37 ileixe np :)
06:26:59 openstackgerrit Adam Harwell proposed openstack/nova master: Add scheduler filter for routed_segments https://review.opendev.org/709280
08:43:36 gibi stephenfin: I made a review round on the series https://review.opendev.org/#/c/689388/ I have some questions and suggestions.
09:00:08 openstackgerrit Balazs Gibizer proposed openstack/nova master: Remove unnecessary parentheses https://review.opendev.org/709652
09:04:31 maysams Hello folks, I saw some discussions few years back regarding creating Nova VM with vsock
09:04:40 maysams but I'm not sure whether the support for it exists as no doc was found
09:04:50 maysams Could you let me know if this is allowed?
09:41:24 openstackgerrit Kevin Zhao proposed openstack/nova master: Add default cpu model for aarch64 https://review.opendev.org/709494
10:41:41 openstackgerrit Stephen Finucane proposed openstack/nova master: mypy: Add type annotations to 'nova.pci' https://review.opendev.org/676209
10:41:41 openstackgerrit Stephen Finucane proposed openstack/nova master: tox: Integrate mypy https://review.opendev.org/676208
10:41:42 openstackgerrit Stephen Finucane proposed openstack/nova master: WIP: mypy: Add type annotations to top-level modules https://review.opendev.org/705658
10:41:42 openstackgerrit Stephen Finucane proposed openstack/nova master: mypy: Add nova.cmd, nova.conf, nova.console https://review.opendev.org/705657
10:53:29 Tony31 Anyone here?
10:56:24 kashyap Tony31: Just ask your question, someone might answer.
10:56:50 kashyap People live in different timezones. So please be patient, if it isn't answered right away.
11:26:33 openstackgerrit Stephen Finucane proposed openstack/nova master: Fix incorrect vm and task state after build failure race https://review.opendev.org/689388
11:26:33 openstackgerrit Stephen Finucane proposed openstack/nova master: Unplug VIFs as part of cleanup of networks https://review.opendev.org/663382
11:48:27 stephenfin bauzas: Easy +2 here, if you're about https://review.opendev.org/#/c/703146/
11:48:41 bauzas I can, children vacations \o/
11:59:30 openstackgerrit Stephen Finucane proposed openstack/nova master: api: Add microversion 2.82, extra spec validation https://review.opendev.org/708436
11:59:30 openstackgerrit Stephen Finucane proposed openstack/nova master: api: Add framework for extra spec validation https://review.opendev.org/704643
12:44:56 sean-k-mooney bauzas: i commented on https://review.opendev.org/#/c/656885/3 with my vision of how the route networking supprot should be done. i 100% think this needs a spec and i think we should not focus on just routed network but instead on network aware schdeuling instead.
12:46:35 sean-k-mooney we have talked about that before at the ptg but its why im opposed to the poc direction as i think it will not be a good fit as it does not adress the generic problem and only the niech usecase of routed network in stead of network aware scheduling which we have disucess at the ptg a few times
14:26:46 kashyap stephenfin: Hiya ... it's that time of the day (rST quirks):
14:27:12 kashyap ``Broadwell``, ``Broadwell-IBRS``, ``Broadwell-noTSX``, ``Broadwell-noTSX-IBRS`` Intel Core Processor (Broadwell, 2014)
14:27:12 kashyap stephenfin: Is there a way to line-break this long-line?
14:27:24 kashyap stephenfin: Err, let me get a pastebin. It's a definition list
14:29:13 kashyap The first line is 80 lines long (I want to wrap it at 72, but the Sphinx parser complains if you wrap the definition list): https://paste.centos.org/view/4235e78b
14:29:48 kashyap I guess there's no way around it.

Earlier   Later