Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-01-15
15:00:26 Sundar sean-k-mooney: You don't need https://review.opendev.org/#/c/696089/ either
15:00:52 sean-k-mooney Sundar: cool i did not have either applied so ill only use the first one you linked
15:01:15 sean-k-mooney https://review.opendev.org/#/c/698846/
15:02:16 dansmith Sundar: replying to those comments is a lot harder 1-2 weeks after I make them, vs sooner
15:02:47 Sundar dansmith: Got it, I had to fix the Cyborg side event nptification patch and its dependencies before I got to this
15:02:56 dansmith Sundar: I'm trying to make an effort to drop whatever I'm doing to review those patches as soon as you put them up to keep things moving, so it'd be helpful if you could at least reply to them quicker, even if the code isn't coming immediately
15:03:44 dansmith I think I've replied to all the ones you're expecting to hear about, but if I missed one let me know
15:04:33 Sundar dansmith: OK, but without trying the code changes, the best I could say is 'will consider/try it out'. Presumably you want a firmer reply than that.
15:04:45 Sundar dansmith: Thanks, looking now
15:05:20 dansmith Sundar: well, some of the comments were not needing code changes. anyway, just saying, faster responses would be nice
15:05:40 Sundar sena-k-mooney: Not sure if 'openstack accelerator device profile' is ready for prime time yet. But here it seems to be fine. Do you have the cyborg API logs?
15:05:53 Sundar sean-k-mooney: ^
15:10:55 sean-k-mooney am i still have it deploy it seamed to create the profile fine as the list shows
15:11:12 sean-k-mooney its a littel weird to just dump the raw json on the command line
15:11:25 sean-k-mooney when defining the groups
15:12:08 sean-k-mooney as far as i can tell the issue was the root_required patch not being present on the branch i tested
15:14:04 sean-k-mooney Sundar: next time i try goign though the work flow of deploying a vm ill see if i can capture the api logs too
15:14:37 gibi stephenfin, efried: the qos live migration patches are ready for review, if you have some time https://review.opendev.org/#/q/status:open++topic:bp/support-move-ops-with-qos-ports-ussuri
15:16:51 Sundar sean-k-mooney: Thanks
15:18:10 Sundar dansmith: Re. https://review.opendev.org/#/c/631244/53/nova/tests/unit/compute/test_compute_mgr.py@6098 , I stepped through the code and noticed that the return value was a dict, not a list. In fact, the first dict from the list.
15:19:02 Sundar dansmith: Your concerns about clarity are probably addressed by having mock_get_arqs.assert_has_calls() and clarifying the comments, right?
15:19:45 dansmith Sundar: then I think something else is going on: https://pastebin.com/7Rr1CVfd
15:21:14 dansmith Sundar: side_effect is the one that behaves the way you're describing (as you know): https://pastebin.com/PHiKTtew
15:22:03 Sundar dansmith: let me take another look and get back
15:22:56 dansmith thanks
15:39:15 openstackgerrit Sylvain Bauza proposed openstack/nova-specs master: Proposes NUMA topology with RPs https://review.opendev.org/552924
15:39:25 bauzas huzzah ^
15:42:14 sean-k-mooney well it nice that you rejoin us with such a simple and non contorviral proposal :P
15:49:48 artom Controviral?
15:50:08 artom I kinda like that. Becomes viral because it's so controversial
15:50:33 sean-k-mooney hehe
15:50:47 sean-k-mooney you figured it out so its close enough
15:51:39 sean-k-mooney but yes i think the decieion we have been making related to mixed cpus makes modeing numa in plamcent more controversial
15:52:06 stephenfin gibi: Done (the QoS live migration series)
15:52:17 gibi stephenfin: thanks a lot!
15:52:53 sean-k-mooney i was less concerned when we used the hw:* syntax but if we go with the placementese syntax then i have deep concerns with granuarl request groups and the group_policy as they exist today
15:55:11 Sundar dansmith: Please see http://paste.openstack.org/show/788413/. The main change I made is http://paste.openstack.org/show/788415/. The first call to get_arqs_for_instance is mocked properly, but the second call hits a StopIteration error.
15:57:38 dansmith Sundar: no, return_value will work as many times as you want: https://pastebin.com/2Wdgb64J
15:58:20 sean-k-mooney if you want to have a limited number of retruns or different retrun i think you use sideffect with an iterable
15:58:42 Sundar Ah, my bad, I still have it as side_effect. Will fix that
15:58:53 dansmith ...
16:01:04 Sundar I think it should work with these changes. This was the only pending point, AFAICS. Are you good with updating the patch set once this is resolved?
16:01:11 Sundar dansmith: ^
16:01:45 dansmith Sundar: unless you've replied to other comments I had, I think so
16:02:56 Sundar I pretty much agreed to other changes, even if it is a matter of preference.
16:03:24 aarents lyarwood: I replied to your remark https://review.opendev.org/#/c/696084 maybe kashyap can confirm that implemeting finish_unshelved as you suggest is the way to go
16:23:03 openstackgerrit Sundar Nadathur proposed openstack/os-traits master: Add a trait that compute manager can handle accelerator requests. https://review.opendev.org/698961
16:43:53 openstackgerrit Balazs Gibizer proposed openstack/nova stable/stein: DNM: try to trace bug 1859766 https://review.opendev.org/702700
16:43:53 openstack bug 1859766 in OpenStack Compute (nova) stein "functional tests intermittently fails with "ReadOnlyFieldError: Cannot modify readonly field uuid"" [Undecided,New] https://launchpad.net/bugs/1859766 - Assigned to Balazs Gibizer (balazs-gibizer)
17:27:53 openstackgerrit Kashyap Chamarthy proposed openstack/nova-specs master: Re-propose "Secure Boot support for KVM & QEMU guests" for Ussuri https://review.opendev.org/693844
17:28:41 kashyap aarents: Sorry, I wasn't ignoring that change, just been buried.
17:46:06 stephenfin efried_afk: not sure if you're gone for the day, but https://review.opendev.org/#/c/697537/ could probably be revisited now that x-cell resize has merged
17:47:03 stephenfin and melwitt, I think I've clarified my (new) intent in https://review.opendev.org/#/c/696745/, which should assuage some of your concerns hopefully
17:47:08 stephenfin with that
17:47:32 melwitt stephenfin: ack
18:02:52 openstackgerrit Artom Lifshitz proposed openstack/nova master: NUMA LM: Add func test for bug 1845146 https://review.opendev.org/687404
18:02:52 openstack bug 1845146 in OpenStack Compute (nova) train "NUMA aware live migration failed when vCPU pin set" [High,Fix committed] https://launchpad.net/bugs/1845146 - Assigned to Dan Smith (danms)
18:03:46 artom dansmith, could I ask you to look at that 2-patch series ^^ when you get a chance? stephenfin's the other reviewer, but your +1 or +W'less +2 can still carry weight and help the evential non-RH approver (efried, I'm guessing)
18:12:05 dansmith artom: yeah, although I'm a couple layers deep on the stack and have a call in 18 minutes
18:12:27 dansmith artom: meaning, don't let me forget
18:14:15 fungi anybody remember how far back nova started providing configdrive by default without needing to manually enable it in the config?
18:14:29 fungi (i mean providing it if the client requests it at boot)
18:14:35 dansmith a long time
18:15:06 fungi cool, config reference only goes back as far as pike but seemed to indicate it was available by default at least that far back
18:15:10 fungi thanks dansmith!
18:15:27 dansmith yeah I'm pretty sure it's been around for a real (real) long time
18:15:27 sean-k-mooney fungi: its been availabel via the api
18:15:55 sean-k-mooney its not on by defualt unless you request it or set the force config drive options in the config
18:16:12 sean-k-mooney it might predate micorverions
18:16:23 fungi yep, thanks. trying to help a user in #zuul figure out whether they have to do anything fancy in nova's config to make configdrive available to nodepool
18:16:25 dansmith it definitely predates microversions
18:16:42 fungi sounds like no, just make sure nodepool asks for a configdrive
18:17:03 dansmith well, isn't there a way to configure it to *not* honor the user's configdrive request?
18:17:47 dansmith ah, nevermind, looks like they can always ask for it
18:17:48 dansmith so yeah I think you're good
18:18:45 sean-k-mooney fungi: the only thing they shoudl have to do is make sure the iso vfat tools are installed so nova can actully create the config drive
18:18:57 sean-k-mooney but they will be on any normal install
18:19:14 fungi right, already suggested that to them
18:19:24 fungi either genisofs or mkisofs depending on the vintage
18:19:36 sean-k-mooney yep
18:19:49 fungi i have a feeling they just need to adjust their nodepool launcher config
18:20:58 fungi they provided a `openstack show server` paste which indicated config_drive was not True and i expect that comes down to the boot parameters it provided
18:22:22 sean-k-mooney right. it could be forced in there config but nodepool has a setting for that too right
18:23:01 sean-k-mooney you can use https://zuul-ci.org/docs/nodepool/configuration.html#attr-providers.[openstack].diskimages.config-drive
18:24:04 sean-k-mooney actully it looks like node pool does not use the boot parmater since that is on the image and instead is using the image metadata right?
18:24:57 sean-k-mooney there are 3 ways to trun on cofig drive, the force_config_drive config option, the api boot request or the image metadata
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: ksa auth conf and client for Cyborg access https://review.opendev.org/631242
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Add Cyborg device profile groups to request spec. https://review.opendev.org/631243
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Define Cyborg ARQ binding notification event. https://review.opendev.org/692707
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Create and bind Cyborg ARQs. https://review.opendev.org/631244
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Pass accelerator requests to each virt driver from compute manager. https://review.opendev.org/698581
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Compose accelerator PCI devices into domain XML in libvirt driver. https://review.opendev.org/631245
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Delete ARQs for an instance when the instance is deleted. https://review.opendev.org/673735
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Enable hard/soft reboot with accelerators. https://review.opendev.org/697940
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Enable start/stop of instances with accelerators. https://review.opendev.org/699553
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Enable and use COMPUTE_ACCELERATORS trait. https://review.opendev.org/699554
18:31:46 openstackgerrit Sundar Nadathur proposed openstack/nova master: Add cyborg tempest job. https://review.opendev.org/670999
18:39:56 openstackgerrit waleed mousa proposed openstack/os-vif master: [Follow Up] OVS DPDK port representors support https://review.opendev.org/665965
18:43:29 artom dansmith, sorry, was on a 1:1
18:43:36 artom dansmith, thanks! I'll be sure to hound your ass then

Earlier   Later