| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-06-19 | |||
| 13:22:20 | mriedem | when you set a port binding resource to active, it atomically updates the port's binding:host_id field | |
| 13:22:27 | mriedem | it's confusing that they are different, but that's the way it is | |
| 13:22:52 | mriedem | re: https://review.openstack.org/#/c/523604/18/nova/network/neutronv2/api.py@1223 | |
| 13:22:59 | gibi | Ahh I see now. the difference between the spec and the neutron code did not make this easier :) | |
| 13:23:12 | mriedem | yeah as i said in ^ the spec is outdated | |
| 13:23:16 | gibi | yeah | |
| 13:23:27 | mriedem | and the neutron ml2 plugin and port binding extended api reference doesn't yet have api ref docs | |
| 13:23:33 | mriedem | something i asked mlavalle about | |
| 13:24:28 | mriedem | note that if the binding resource field was 'host_id' rather than 'host' the live migration job would be failing at the top of the series | |
| 13:25:00 | mriedem | hmm, which it is now, /me looks | |
| 13:25:49 | mriedem | haven't seen this before http://logs.openstack.org/37/522537/29/check/nova-live-migration/08534a4/logs/screen-n-cpu.txt.gz?level=TRACE#_Jun_19_01_11_49_573713 | |
| 13:26:37 | mriedem | this might be something new in libvirt 4.0.0 using the queens UCA | |
| 13:26:43 | mriedem | which we weren't using last time i checked these job results | |
| 13:28:38 | gibi | having a tempest test at the top helps, thanks | |
| 13:29:47 | gibi | mriedem: regarding the rollback of multiple port activations. I don't know what can we do so let's keep it as is and try to fix it later when we see the real problem | |
| 13:30:08 | gibi | mriedem: I will try to get back to your neutron-new-port-binding-api series to re-review but it might not be today | |
| 13:30:48 | mriedem | gibi: sure, that's ok, thanks for what you've done so far | |
| 13:37:23 | gibi | mriedem: btw, Kevin_Zheng fixed up https://review.openstack.org/#/c/564092/ and I'm +2 so you can send it in | |
| 13:37:48 | mriedem | ok | |
| 13:40:01 | gibi | mriedem: I tried to reporduce bug 1777540 but I failed at https://bugs.launchpad.net/nova/+bug/1777157/comments/4 | |
| 13:40:03 | openstack | Launchpad bug 1777157 in OpenStack Compute (nova) "cold migration fails for ceph volume instances" [Undecided,New] | |
| 13:40:04 | openstack | bug 1777540 in OpenStack Compute (nova) ""IndexError: list index out of range" in ExceptionPayload.from_exception during resize failure" [Undecided,New] https://launchpad.net/bugs/1777540 | |
| 13:40:40 | gibi | mriedem: I don't know how can I hit code https://github.com/openstack/nova/blob/530c24596a74a6115f2e944176c46e4f3d199811/nova/compute/manager.py#L4079 | |
| 13:40:47 | gibi | mriedem: from the nova API | |
| 13:42:14 | gibi | mriedem: is short how can I trigger a migrate / resize that targets the same host and does not change the flavor of the instance | |
| 13:44:44 | mriedem | gibi: you could specify the host field when cold migrating? | |
| 13:45:00 | mriedem | https://developer.openstack.org/api-ref/compute/#migrate-server-migrate-action | |
| 13:45:29 | gibi | mriedem: I tried, I got {"badRequest": {"code": 400, "message": "Cannot migrate to the host where the server exists."}} | |
| 13:45:46 | mriedem | hmm, because https://github.com/openstack/nova/blob/530c24596a74a6115f2e944176c46e4f3d199811/nova/compute/api.py#L3366 | |
| 13:46:14 | mriedem | which is technically a behavior change since some drivers allow you to migrate to self https://github.com/openstack/nova/blob/530c24596a74a6115f2e944176c46e4f3d199811/nova/compute/manager.py#L4077 | |
| 13:46:21 | mriedem | thinking vmware | |
| 13:46:59 | openstackgerrit | Dan Smith proposed openstack/nova stable/queens: [Stable Only] Add amd-ssbd and amd-no-ssb CPU flags https://review.openstack.org/576270 | |
| 13:46:59 | mriedem | yeah vmware is the only driver that has supports_migrate_to_same_host=True | |
| 13:47:11 | mriedem | since the compute host is just managing a vcenter cluster of esxi hosts | |
| 13:47:50 | gibi | mriedem: it is added in Queens by https://review.openstack.org/#/c/408955/ | |
| 13:48:32 | mriedem | yeah i remember reviewing it :( | |
| 13:48:57 | gibi | mriedem: but the bug report says this issue was visible in queens so I don't get it how | |
| 13:49:10 | mriedem | so they weren't picking a host for the cold migrate, | |
| 13:49:15 | mriedem | maybe they only have 1 compute? | |
| 13:49:23 | gibi | then I get NoValidHost | |
| 13:49:38 | mriedem | or, maybe aggregates are setup in such a way that the same host is picked? | |
| 13:50:09 | mriedem | or you have 2 hosts and the other is full or disabled? | |
| 13:50:26 | gibi | OK, good ideas, let me dig | |
| 13:51:04 | mriedem | gibi: you also have to have allow_resize_to_same_host=True in nova.conf | |
| 13:53:37 | sahid | mriedem: thanks will look at that | |
| 13:53:59 | gibi | mriedem: does allow_resize_to_same_host affects migration? | |
| 13:56:31 | gibi | mriedem: yes, it does | |
| 13:56:55 | openstackgerrit | Dan Smith proposed openstack/nova stable/queens: [Stable Only] Add amd-ssb and amd-no-ssb CPU flags https://review.openstack.org/576270 | |
| 13:58:27 | dansmith | kashyap: okay, I'm confused from all your edits to the bug.. going off of the libvirt patch it says it should be amd-ssbd and amd-no-ssb | |
| 13:59:47 | dansmith | I just processed the email notifications from your bug edits and it seems like you dropped the 'b' off the end of the first | |
| 13:59:49 | dansmith | er, the 'd' | |
| 14:01:54 | openstackgerrit | Dan Smith proposed openstack/nova stable/queens: [Stable Only] Add amd-ssbd and amd-no-ssb CPU flags https://review.openstack.org/576270 | |
| 14:01:56 | dansmith | whatever, changing it to match the libvirt patch | |
| 14:02:04 | dansmith | changing it *back* I should say | |
| 14:03:01 | kashyap | dansmith: Damn, I was bleary-eyed without coffee in the morning, and mistook Matt's comment | |
| 14:03:06 | kashyap | My bad; let me go update it right away | |
| 14:03:07 | dansmith | the bug also still calls it "amdb-no-ssb" by the way | |
| 14:03:15 | kashyap | Sorry about that. | |
| 14:03:19 | kashyap | Adjusting it | |
| 14:04:18 | kashyap | Originally I typoed only one place, in the summary. But somehow thought I did it all over the description. | |
| 14:04:45 | dansmith | and I copied those two names into the conf file, but then used the libvirt patch as reference for all the text I typed | |
| 14:07:06 | bauzas | efried: good morning | |
| 14:07:14 | efried | bauzas: Howdy! | |
| 14:07:37 | bauzas | efried: about efried: about https://review.openstack.org/#/c/564214/5/nova/tests/unit/virt/libvirt/test_driver.py@19853 the problem is that I need to provide a nova.conf fake file for this | |
| 14:07:45 | mriedem | alex_xu: so should we put https://review.openstack.org/#/c/560317/ in a runway or hold off for now? | |
| 14:08:14 | bauzas | efried: I can't just set the options directly ;) | |
| 14:08:31 | efried | bauzas: whyzat? Because the groups are dynamic? | |
| 14:09:07 | efried | oh, sorry, I was thinking of the wrong patch. | |
| 14:09:47 | efried | bauzas: wait, no I'm not. Can't you set the one opt, then call the dynamic conf thingy, then set the subsequent opt? | |
| 14:10:22 | kashyap | dansmith: Updated the bug. Struggling with too many acronyms. :-) | |
| 14:12:41 | bauzas | efried: well, if I'm providing 2 types but only set one opt, the second opt would get an exception | |
| 14:12:59 | mriedem | cdent: i think i'm going to put the optional placement db series into the available runway slot, are you around to deal with reviews on it for the next 2 weeks (shouldn't really take that long, the big pieces have merged it looks like) | |
| 14:13:15 | openstackgerrit | Merged openstack/os-traits master: Adds HW_CPU_HYPERTHREADING standard trait https://review.openstack.org/576030 | |
| 14:13:33 | cdent | a) cool, b) yes, c) there was a big bug in the next in line, but I fixed it, so we good | |
| 14:13:45 | mriedem | ack | |
| 14:13:47 | cdent | I except there will be some documentation concerns, which is good and right | |
| 14:13:51 | efried | bauzas: Sorry, I'm not understanding. Why wouldn't you provide 2 types and set both opts? | |
| 14:14:16 | bauzas | efried: sec | |
| 14:14:58 | sq4ind | Hey guys, have a problem with long running tasks (migrations). I've enabled send_service_user_token and allow_expired but still sometimes it fails. Here are logs: https://pastebin.com/cY82PP34 . Any idea? | |
| 14:15:01 | sq4ind | Thanks | |
| 14:15:53 | bauzas | efried: something like this ? http://paste.openstack.org/show/723830/ | |
| 14:16:14 | bauzas | efried: if so, looks a bit weird | |
| 14:16:30 | efried | bauzas: sec... | |
| 14:17:11 | bauzas | efried: oops, of course I meant http://paste.openstack.org/show/723831/ | |
| 14:18:27 | efried | bauzas: Does self.flags not let you set an opt that's not registered? | |
| 14:18:40 | bauzas | efried: unfortunately, no | |
| 14:18:44 | efried | boo. | |
| 14:18:49 | dansmith | or, fortunately. | |
| 14:18:55 | bauzas | heh | |
| 14:19:35 | bauzas | efried: self.flags is like conf.override | |
| 14:19:42 | bauzas | you need to have an existing group | |
| 14:19:57 | bauzas | actually, sec | |
| 14:20:46 | bauzas | efried: I think I can test it | |
| 14:22:04 | efried | bauzas: okay; the other option is to split apart _configure_gpu_mapping, one bit to register the groups and one to grab the values and do the validation. Then you can call the former from your test before you set self.flags; then call the latter. | |
| 14:22:39 | bauzas | efried: I think I can mock register_opts | |
| 14:24:18 | openstackgerrit | Lee Yarwood proposed openstack/nova master: Add regression test for bug #1764883 https://review.openstack.org/562072 | |
| 14:24:19 | openstack | bug 1764883 in OpenStack Compute (nova) "Evacuation fails if the source host returns while the migration is still in progress" [Undecided,In progress] https://launchpad.net/bugs/1764883 - Assigned to Lee Yarwood (lyarwood) | |
| 14:24:19 | openstackgerrit | Lee Yarwood proposed openstack/nova master: compute: Ensure pre-migrating instances are destroyed during init_host https://review.openstack.org/562284 | |
| 14:30:58 | openstackgerrit | Lee Yarwood proposed openstack/nova master: compute: Ensure pre-migrating instances are destroyed during init_host https://review.openstack.org/562284 | |
| 14:40:16 | mriedem | this should be an easy +W for someone to fix busted compat tests https://review.openstack.org/#/c/576256/ | |