Earlier  
Posted Nick Remark
#openstack-nova - 2019-01-11
12:27:43 lyarwood stephenfin: ack sure thing
13:07:43 jaypipes line.
13:07:43 jaypipes stephenfin, bauzas, sean-k-mooney: I'm fed up with the cpu resource tracking spec. It's just analysis paralysis at this point, and I'm tired of retreading the same points about CONF.vcpu_pin_set over and over. Since at this point, folks from Intel and RH are really the only ones that seem to understand how any of the NUMA and CPU topology stuff "should" work, it's probably best to have an Intel or RH employee try to get this over the finish
13:08:10 jaypipes I've essentially given up trying to refactor any of the poo in Nova any more.
13:08:21 jaypipes just seems like a completely wasted effort on my part.
13:11:54 cdent it does seem a bit paralyzed
13:12:16 kashyap jaypipes: Mind posting the spec link? Your comment makes me curious
13:14:01 cdent kashyap: https://review.openstack.org/#/c/555081/
13:15:45 kashyap Thanks
13:16:40 kashyap Oh dear, look at that scrollbar.
14:11:40 stephenfin jaypipes: Yeah, that's fair. The thought that maybe we'd be better of debating this stuff during code did cross my mind yesterday, but I'd been hoping to agree that on the spec
14:12:25 stephenfin jaypipes: As-is, I'm not sure how to make it any easier. It is a load of poo, as you rightly point out, and the nest of dependencies is somewhat ridiculous
14:13:04 openstackgerrit Paul-Emile Element proposed openstack/os-traits master: Add new traits for TPM support https://review.openstack.org/629299
14:14:17 mnaser good morning everyone. i need to rename a bunch of hostnames in a cloud, which can go very bad in the past
14:14:29 mnaser do we have a documented procedure or should i just evacuate hosts and reboot them with a new host and call it a day
14:25:42 jaypipes mnaser: no idea what the best practice is but will eagerly consume others' suggestions.
14:27:39 mnaser because the hypervisor host seems to be engrained in so many places like instances table, placement db
14:27:54 mnaser and i know before if the hostname changed and nova-compute restarted it would self destruct :)
14:29:33 cdent mnaser: I think unless you do some hand holding, placement will with orphaned inventory, throwing things out of wack. I don't think that has changed. But it is hard to keep track
14:30:08 mnaser i think i'll go the evacuate host/change hostname/reboot route
14:30:13 mnaser slower but probably reliable i guess
14:35:26 openstackgerrit Merged openstack/os-traits master: Add new traits for TPM support https://review.openstack.org/629299
14:44:36 temka stephenfin, docmaster, an easy review for you: https://review.openstack.org/#/c/629904/
14:45:39 finucannot temka: thanks for the reminder
14:46:23 finucannot temka: done
14:46:31 temka finucannot, cheers!
14:48:29 finucannot jaypipes: If you're not totally burnt by NUMA'y stuff, I'd appreciate some eyes on this bugfix https://review.openstack.org/#/c/629281/
14:53:56 cdent dansmith, jaypipes, gibi, melwitt: does wednesday at 5utc seems a reasonable time to have the placement check in meeting? Or should we wait until monday and mriedem being back to pick a real time
14:56:14 dansmith cdent: do you mean 1700UTC?
14:56:38 cdent dansmith: yes
14:56:45 cdent sorry for being ambiguos
14:57:42 dansmith works for me time wise, but obviously not sure what matt might have on his calendar
14:59:12 cdent i'll suggest wednesday in the pupdate, and we can firm up the actual time on monday
14:59:13 gibi cdent: works for me
14:59:40 cdent gibi: the bandwith thing is on tuesday, yeah?
14:59:46 gibi cdent: the bandwidth hangouts is currently scheduled for tuesday 17:00 utc, but I got feedback from rubasov that it does not work for him, so I thinking about waiting for mriedem's input before move the bandwidth hangouts again.
14:59:57 cdent ah
15:00:10 gibi the joy of finding a good timeslot :)
15:20:14 melwitt finucannot: yeah, wow, I had no idea gantt was still around in gerrit etc. feel free to email
15:22:09 melwitt cdent: sounds good, thanks
16:11:45 devfaz Hi, is there a easy way to (re)create the tap-devices of running instances if the openvswitch.db got cleared? I already restarted openvswitch-agent and this restored all other interfaces, just the tap-devices for nova are missing?
16:23:21 devfaz so found a way "ip link del qvb-XXX" and restart openvswitch-agent. Is this a "good" workaround or is there a better way to restore the network if ovs lost its settings?
16:43:32 openstackgerrit Merged openstack/nova master: Doc: rebuild can result in SHUTOFF VM state https://review.openstack.org/629904
16:47:07 finucannot melwitt: Sweet, done. Guess we need bauzas or johnthetubaguy to unilaterally merge these patches since they're the only active folks in gantt-core https://review.openstack.org/#/c/630154/ https://review.openstack.org/#/c/630138/
16:47:45 melwitt ack, thanks
16:56:54 gibi kashyap: I have some nits in https://review.openstack.org/#/c/629627 If you ack that you can do them in the followup then can +W that patch
16:59:40 openstackgerrit Jack Ding proposed openstack/nova master: [WIP] Flavor extra spec and image properties validation https://review.openstack.org/620706
17:50:58 openstackgerrit Jack Ding proposed openstack/nova master: Correct instance port binding for rebuilds https://review.openstack.org/603844
18:20:59 openstackgerrit Andriy Shevchenko proposed openstack/gantt master: Update home-page https://review.openstack.org/630096
18:50:51 sean-k-mooney devfaz if you just restated nova-compute it would replug teh vm interfaces automatically on startup which would repopulate teh ovs-db
22:25:37 openstackgerrit Eric Fried proposed openstack/nova master: Consolidate inventory refresh https://review.openstack.org/615695
22:25:37 openstackgerrit Eric Fried proposed openstack/nova master: Reduce calls to placement from _ensure https://review.openstack.org/615677
22:25:38 openstackgerrit Eric Fried proposed openstack/nova master: Rip the report client out of SchedulerClient https://review.openstack.org/617042
22:25:38 openstackgerrit Eric Fried proposed openstack/nova master: Commonize _update code path https://review.openstack.org/615705
22:25:39 openstackgerrit Eric Fried proposed openstack/nova master: Use a static resource tracker in compute manager https://review.openstack.org/620711
22:25:39 openstackgerrit Eric Fried proposed openstack/nova master: Rip out the SchedulerClient https://review.openstack.org/617049
23:13:35 openstackgerrit Eric Fried proposed openstack/nova master: Reduce calls to placement from _ensure https://review.openstack.org/615677
23:13:36 openstackgerrit Eric Fried proposed openstack/nova master: Commonize _update code path https://review.openstack.org/615705
23:13:36 openstackgerrit Eric Fried proposed openstack/nova master: Consolidate inventory refresh https://review.openstack.org/615695
23:13:37 openstackgerrit Eric Fried proposed openstack/nova master: Rip out the SchedulerClient https://review.openstack.org/617049
23:13:37 openstackgerrit Eric Fried proposed openstack/nova master: Rip the report client out of SchedulerClient https://review.openstack.org/617042
23:13:38 openstackgerrit Eric Fried proposed openstack/nova master: Use a static resource tracker in compute manager https://review.openstack.org/620711
23:15:00 fried_rice jaypipes: This series ought to be ready again now ^
23:16:12 fried_rice Caveat: I haven't gotten to a point where I can tox locally on my shiny new (Windows) system, and don't yet have other hardware on which to install Linux, so I didn't get to do any local testing.
#openstack-nova - 2019-01-12
19:28:53 openstackgerrit Eric Fried proposed openstack/nova master: Consolidate inventory refresh https://review.openstack.org/615695
19:28:53 openstackgerrit Eric Fried proposed openstack/nova master: Reduce calls to placement from _ensure https://review.openstack.org/615677
19:28:54 openstackgerrit Eric Fried proposed openstack/nova master: Rip the report client out of SchedulerClient https://review.openstack.org/617042
19:28:54 openstackgerrit Eric Fried proposed openstack/nova master: Commonize _update code path https://review.openstack.org/615705
19:28:55 openstackgerrit Eric Fried proposed openstack/nova master: Use a static resource tracker in compute manager https://review.openstack.org/620711
19:28:55 openstackgerrit Eric Fried proposed openstack/nova master: Rip out the SchedulerClient https://review.openstack.org/617049
19:29:23 efried1 that oughtta fix the unit tests, silly mistake.
21:48:19 openstackgerrit Eric Fried proposed openstack/nova master: Turn off rp association refresh in nova-next https://review.openstack.org/616033
21:59:51 openstackgerrit Eric Fried proposed openstack/nova master: Use a static resource tracker in compute manager https://review.openstack.org/620711
#openstack-nova - 2019-01-13
03:37:48 openstackgerrit Eric Fried proposed openstack/nova master: Use a static resource tracker in compute manager https://review.openstack.org/620711
09:22:14 openstackgerrit garyk proposed openstack/nova master: Enhance exception raiased when invalid power state https://review.openstack.org/630496
14:53:59 openstackgerrit garyk proposed openstack/nova master: Enhance exception raised when invalid power state https://review.openstack.org/630496
#openstack-nova - 2019-01-14
00:19:30 eandersson What kind of impact does getting additional filtering / weighting data have on scheduler performance?
00:19:54 eandersson e.g. flavor information
00:20:59 eandersson It's easy enough to test with a few computes and VMs, but not sure how to test it on more realistic PROD scales.
07:44:22 openstackgerrit Zhenyu Zheng proposed openstack/nova master: Allow run metadata api per cell https://review.openstack.org/624612
08:05:56 maciejjozefczyk jaypipes: hey. I did a little update on counters here: https://review.openstack.org/#/c/614167/ Please check that ;) Sorry for spamming a lot about this
11:48:59 finucannot bauzas: If you're about today, fancy pushing https://review.openstack.org/630154 and https://review.openstack.org/630138 ?
12:35:36 jaypipes maciejjozefczyk: no worries! will look at it shortly.
13:28:22 sean-k-mooney o/
13:51:24 gibi mriedem: hi! welcome back! I'm trying to find a slot for a hangouts call about the bandwidth patches and need your input. Possible slots this week: Wednesday 18:00 UTC (after cdent's placement hangouts), Thursday 17:00 UTC, Friday 17:00 UTC
13:51:26 openstackgerrit Matthew Booth proposed openstack/nova master: Fix configure() called after DatabaseAtVersion fixture https://review.openstack.org/619723
13:53:40 gibi mriedem: and one more not so good spot on Tuesday 17:00 UTC
13:55:07 mriedem any of those should work for me
13:56:47 efried n-sch meeting in 10 minutes in #openstack-meeting-alt
14:00:07 gibi mriedem: thanks
14:00:14 gibi mriedem: will post the final time on the ML
14:05:14 openstackgerrit caoyuan proposed openstack/nova stable/rocky: Update the description to make it more accuracy https://review.openstack.org/630681
14:33:54 efried belmoreira: Howdy. Are you around today?
14:39:29 asmita gibi: Hi
14:41:17 gibi asmita: hu
14:41:20 gibi asmita: hi
14:42:25 sean-k-mooney adrianc: do you mind if i rebase the sriov patch series and start addressing some of stephenfin's nits
14:43:13 asmita gibi: How do we use this mehod?[1]:https://github.com/openstack/python-novaclient/blob/master/novaclient/tests/unit/utils.py#L30
14:43:27 asmita method*
14:44:22 sean-k-mooney mriedem: are you ok with removing your procedual -2 in https://review.openstack.org/#/c/616120/ since the spec has been merged and blueprint approved?

Earlier   Later