| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-08-29 | |||
| 14:32:19 | kosamara | efried: I had looked at Cyborg in spring, but haven't been following it since. The primary use case that I saw Cyborg enabling was FPGA function-aaS. | |
| 14:32:51 | mriedem | efried: i don't know what is pushing the initial inventory data | |
| 14:32:54 | mriedem | so i can't really say | |
| 14:32:54 | efried | kosamara: Yes, that's kind of the initial motivator, but the project is supposed to subsume all device management. | |
| 14:32:56 | mriedem | we don't log anything | |
| 14:33:28 | mriedem | so i'm going to push some debug patches so we can .... debug | |
| 14:33:43 | mriedem | my guess is the local provider tree cache is current but remote is not, and we never update | |
| 14:33:43 | efried | ack, let me know if I can help. | |
| 14:39:50 | efried | mriedem: Well, looking at has_inventory_changed, it *should* be paying attention to alloc ratio updates, whether the field exists and is being changed, or doesn't exist and is being added. | |
| 14:40:10 | efried | and has_inventory_changed being false is the only way we would avoid sending the update down to placement. | |
| 14:40:33 | efried | ...barring actual errors which we would see in the compute log | |
| 14:43:43 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Log the operation when updating generation in ProviderTree https://review.openstack.org/597553 | |
| 14:44:10 | efried | mriedem: FYI: http://paste.openstack.org/show/729053/ <== looks fine. | |
| 14:47:08 | kosamara | efried: Does that mean that instead of proposing improvements to the way Nova manages devices, we should contribute this functionality to Cyborg, because in the future Cyborg will have exclusive responsibility in that domain? | |
| 14:47:24 | efried | kosamara: Well, that's what needs to be discussed, really. | |
| 14:47:51 | efried | I am pretty far behind on the massive volume of cyborg specs that are out there | |
| 14:48:05 | efried | but I haven't yet seen one where they deal with discovery and whitelisting. | |
| 14:48:07 | kosamara | efried: I was going to ask if you have any pointers to similar work there | |
| 14:48:25 | efried | It's entirely likely that they've got that proposed and I just haven't gotten to it yet. | |
| 14:48:26 | kosamara | and modelling in Placement with RPs? | |
| 14:48:34 | efried | oh, yes, that's definitely their plan. | |
| 14:49:36 | efried | I'm going to try to ask about it in #openstack-cyborg, if you'd like to join me there. | |
| 14:49:42 | kosamara | But can Cyborg create RPs now, or it has to happen through nova, like neutron does in the network bandwidth providers spec? | |
| 14:49:48 | kosamara | ok | |
| 14:50:42 | efried | kosamara: Yeah, that's the question. The providers are intended to be created and "owned" by cyborg code, but I'm still not 100% clear whether that's at the behest/prompting of a nova flow or totally independent. | |
| 14:51:06 | efried | kosamara: Because obviously somebody has to coordinate those device RPs being parented to the compute node RP. | |
| 14:54:45 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Making instance listing skipping down cells configurable https://review.openstack.org/592428 | |
| 14:56:31 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add debug logs for when provider inventory changes https://review.openstack.org/597560 | |
| 14:56:51 | mriedem | naichuans: efried: cdent: jaypipes: zigo: ^ hopefully we can learn something from this | |
| 14:57:53 | mriedem | zigo: i updated your patch with a depends-on to that nova change | |
| 14:57:54 | efried | mriedem: dig | |
| 14:57:57 | cdent | mriedem: do we have a way of setting our test environments so they are more like what zigo and naichuans were experiencing? because apparently our test environments are configuring too much to reflect reality? | |
| 14:58:20 | mriedem | cdent: xenserver ci uses devstack | |
| 14:58:30 | mriedem | it's pretty stock outside of saying use the xen driver rather than libvirt | |
| 14:58:36 | zigo | mriedem: cdent: Would it help if I added these patches to my package and re-run puppet? | |
| 14:58:45 | mriedem | zigo's is using puppet and non-ubuntu but with libvirt | |
| 14:59:01 | zigo | non-ubuntu: Debian Stretch ... :P | |
| 14:59:01 | mriedem | zigo: so my depends-on nova change won't get pulled into that CI run? | |
| 14:59:14 | cdent | mriedem: I get that. The root of my question is: How come we didn't fail tempest or functional? | |
| 14:59:25 | mriedem | cdent: well that's what i'm trying to figure out... | |
| 14:59:38 | zigo | deb http://stretch-rocky.debian.net/debian stretch-rocky-backports main + deb http://stretch-rocky.debian.net/debian stretch-rocky-backports-nochange main | |
| 14:59:39 | mriedem | we do'nt configure allocation ratios in nova.conf in devstack | |
| 15:00:21 | cdent | I know, I'm not being a dick (at least I hope not), I'm just questioning-out-loud | |
| 15:00:51 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Add scatter-gather-single-cell utility https://review.openstack.org/594947 | |
| 15:01:00 | cdent | it seems that if we know what the difference is between the standard gate tests and e.g. the xen tests, we can make our tests fail and work from that | |
| 15:02:06 | mriedem | as far as i know, the xen tests are mostly stock | |
| 15:02:29 | mriedem | they did not hard-code the allocation ratios until jay told them to in the ML as a workaround | |
| 15:02:42 | mriedem | which makes me think, my debug patch probably won't fail the xen ci now b/c of that... | |
| 15:04:48 | openstackgerrit | Merged openstack/osc-placement master: Random names for functional tests https://review.openstack.org/542745 | |
| 15:05:33 | openstackgerrit | Konstantinos Samaras-Tsakiris proposed openstack/nova-specs master: Placement model for passthrough devices https://review.openstack.org/591037 | |
| 15:10:07 | openstackgerrit | Chris Dent proposed openstack/nova master: [placement] Make _ensure_aggregate context not independent https://review.openstack.org/597486 | |
| 15:11:07 | mriedem | anyone remember were the hell the xenserver CI repo is in github? | |
| 15:13:22 | mriedem | guessing https://github.com/citrix-openstack/qa | |
| 15:22:16 | jaypipes | mriedem: sorry? | |
| 15:24:41 | openstackgerrit | Dan Smith proposed openstack/nova master: DNM: Tester for grenade job https://review.openstack.org/597566 | |
| 15:30:16 | openstackgerrit | Chris Dent proposed openstack/nova master: [placement] Make _ensure_aggregate context not independent https://review.openstack.org/597486 | |
| 15:32:04 | dansmith | mriedem: tssurya: melwitt: do we need a cells meeting? my patches merged, I think we got melwitt's from last week as well, and tssurya and I are crushing the down cell stuff | |
| 15:32:20 | dansmith | and when I say we, I mean tssurya is doing it and I'm throwing tomatoes | |
| 15:32:45 | tssurya | I am okay to skip, sorry about the slow-ness in the spec imple; its because I am a litte stuck with the US visa stuff for the next week | |
| 15:33:19 | tssurya | but it should be settled tomorrow finally! | |
| 15:33:31 | mriedem | i don't really have any news; some info is coming in on the cross-cell migration stuff and i'm trying to probe internally (yes wording intended) on how our product team did this for cascading aka cells v1 | |
| 15:33:51 | dansmith | tssurya: oh so you'll be in denver yeah? | |
| 15:33:57 | tssurya | yes :) | |
| 15:34:01 | mriedem | Kevin_Zheng knows more about what our product team did than i ever will | |
| 15:34:27 | dansmith | tssurya: that's cool, although I'm disappointed you'll be seeing the least nice venue in our fair country :/ | |
| 15:34:49 | tssurya | dansmith: haha really ? its actually gonna be my first time in your country | |
| 15:35:04 | dansmith | tssurya: I know, which is why I'm disappointed :/ | |
| 15:35:05 | mriedem | if i remember kevin's email correctly, tl;dr was that ports and volumes are accessible across cells for them | |
| 15:35:24 | mriedem | colorado is "nice" depending on where you are | |
| 15:35:32 | mriedem | out by the airport is not so much | |
| 15:35:40 | dansmith | yes, colorado is for sure | |
| 15:35:49 | dansmith | but the hotel and area we're in sucks | |
| 15:35:54 | tssurya | heh, I booked into the same venue as the conference | |
| 15:35:56 | tssurya | oops | |
| 15:35:59 | mriedem | everyone did | |
| 15:36:02 | dansmith | unless you like living next to a freeway onramp with a big train | |
| 15:36:05 | dansmith | tssurya: you have to be there, yeah | |
| 15:36:15 | dansmith | tssurya: it'snot close to anything, which is part of the problem | |
| 15:36:21 | mriedem | i just hope you like ihop | |
| 15:36:22 | mriedem | every | |
| 15:36:24 | mriedem | morning | |
| 15:36:31 | dansmith | mriedem: don't think we're not | |
| 15:36:35 | mriedem | oh i know | |
| 15:36:43 | mriedem | my stretch pants are on order | |
| 15:36:47 | dansmith | we will put 20lbs on tssurya before she's gone | |
| 15:36:49 | tssurya | rolf | |
| 15:36:53 | dansmith | her family won't recognize her | |
| 15:37:07 | tssurya | :P that is gonna be hard considering the amount I eat | |
| 15:37:09 | dansmith | tssurya: the denver ptg comes with a complimentary case of type 2 diabetes | |
| 15:37:21 | kashyap | LOL | |
| 15:37:27 | tssurya | :D | |
| 15:37:40 | mriedem | luckily she probably gets free health care in switzerland | |
| 15:37:55 | dansmith | they won't know what to do with her | |
| 15:37:58 | kashyap | Is the venue so bad? I read somethings about a train being noisy apparently last time. | |
| 15:38:04 | dansmith | they don't have the IHOP antidote | |
| 15:38:07 | mriedem | kashyap: that is reportedly fixed | |
| 15:38:12 | kashyap | (And you're in the same venue this time -- assuming that problem is fixed) | |
| 15:38:15 | kashyap | mriedem: Ah, I see. | |
| 15:38:18 | tssurya | you both are making it sound really bad, hopefully you are just kidding | |
| 15:38:22 | dansmith | mriedem: med made it sound like maybe not | |