| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-08-03 | |||
| 20:06:29 | dansmith | I don't think powervm via libvirt works from nova | |
| 20:06:35 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Start using oslo_config.sphinxext https://review.openstack.org/482961 | |
| 20:06:35 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Import installation guide https://review.openstack.org/477488 | |
| 20:06:36 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Import administration guide https://review.openstack.org/477497 | |
| 20:06:36 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Rework README to reflect new doc URLs https://review.openstack.org/480074 | |
| 20:06:37 | openstackgerrit | Matt Riedemann proposed openstack/nova master: policies: Fix Sphinx issues https://review.openstack.org/480516 | |
| 20:06:37 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Start using oslo_policy.sphinxext https://review.openstack.org/479358 | |
| 20:06:38 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: Remove dead files https://review.openstack.org/478470 | |
| 20:06:42 | mriedem | the 3rd party ci says it's working | |
| 20:06:52 | dansmith | link? | |
| 20:10:01 | mriedem | http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/nova/65/398265/34/check/tempest-dsvm-full-xenial/8144094/ | |
| 20:10:05 | mriedem | they even have a grenade job | |
| 20:10:11 | mriedem | http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/nova/65/398265/34/check/grenade-dsvm-xenial/27db95e/ | |
| 20:10:26 | dansmith | that's pkvm == powerkvm right? | |
| 20:10:34 | mriedem | yeah | |
| 20:10:43 | mriedem | oh you said powervm via libvirt | |
| 20:10:50 | dansmith | yeah | |
| 20:10:55 | mriedem | yeah nvm then | |
| 20:10:58 | mmedvede | so, it is KVM on Power, not to be confused with PowerKVM OS | |
| 20:11:08 | mmedvede | naming is unfortunate | |
| 20:11:29 | mriedem | https://libvirt.org/drvphyp.html | |
| 20:11:47 | mriedem | is http://www-03.ibm.com/systems/power/software/virtualization/index.html the old powervm driver? | |
| 20:11:59 | mriedem | that ran ssh commands through the vios shell or whatever? | |
| 20:12:04 | mriedem | it's been awhile | |
| 20:12:10 | dansmith | mriedem: phyp via libvirt is the thing I doubt works with nova | |
| 20:12:14 | dansmith | mriedem: which is different than the powervm driver that was in nova at first | |
| 20:12:28 | dansmith | I think the phyp driver uses local hypercall access, IIRC | |
| 20:13:02 | mriedem | hypercall? | |
| 20:13:06 | mriedem | that sounds enterprise ready | |
| 20:13:29 | dansmith | hypercal is like a syscall but to the hypervisor | |
| 20:13:31 | dansmith | *hypercall | |
| 20:14:38 | efried | http://www-03.ibm.com/systems/power/software/virtualization/index.html is the PowerVM that we're enabling via nova.virt.powervm | |
| 20:40:15 | openstackgerrit | Sean Dague proposed openstack/nova master: Add Contributor Guide section page https://review.openstack.org/490644 | |
| 20:45:04 | mriedem | ooo team meeting in 15 minutes | |
| 20:47:40 | sdague | mriedem: https://review.openstack.org/#/c/477488 - follow on comment about why it's fine to have it where it is | |
| 20:48:13 | melwitt | mriedem, dansmith: question on the cellsv2 doc https://review.openstack.org/490612 | |
| 20:48:49 | sdague | mriedem: basically to make the TOC not complete sidebar vomit, we're going to strip a lot out of it an the front page and build topic landing pages with deep links like - https://review.openstack.org/#/c/490644/ | |
| 20:49:23 | sdague | that lets us tidy things up without another giant page shuffle that we'd be 404 chasing | |
| 20:49:39 | mriedem | melwitt: replied | |
| 20:50:09 | openstackgerrit | Merged openstack/nova master: Sum allocations in the scheduler when resizing to the same host https://review.openstack.org/490085 | |
| 20:52:10 | dansmith | mriedem: do you remember the name of the affinity reporting conf item? I was going to add that in there for her, but I can never remember the name | |
| 20:52:21 | melwitt | mriedem: okay, so I guess that means there's nothing wrong with compute node reporting, it's just that the lag between the two (report vs affinity check) could cause a problem with parallel requests | |
| 20:52:46 | dansmith | melwitt: no | |
| 20:52:56 | dansmith | melwitt: there is a compute->scheduler reporting mechanism that can't happen either | |
| 20:53:09 | dansmith | track_instance_changes | |
| 20:53:18 | mriedem | yeah that one | |
| 20:53:19 | dansmith | he had it in his previous comment | |
| 20:53:21 | mriedem | i mentioned that eralier | |
| 20:53:24 | mriedem | *earlier even | |
| 20:53:37 | melwitt | okay, maybe I'm hard of reading :( | |
| 20:53:38 | mriedem | there are about 10 pieces to any single affinity filter puzzle | |
| 20:54:07 | mriedem | like, (1) server group (2) scheduler hint (3) filters (4) various config options | |
| 20:54:14 | mriedem | and then, maybe, it might work | |
| 20:54:18 | dansmith | (5) luck | |
| 20:55:13 | melwitt | yeah, it's just when I read "instance info reporting doesn't work" seemed sort of alarming. the second paragraph explains what the end point is, it's just after reading something like "compute node reporting doesn't work" seems sort of alarming, at least to me | |
| 20:55:18 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Enable cold migration with target host(1/2) https://review.openstack.org/408955 | |
| 20:55:30 | cfriesen | dansmith: /s/luck/timing | |
| 20:57:24 | mriedem | melwitt: dansmith: i've thought about putting some details about the issue into the FAQs page, like "why does affinity/anti-affinity seem to suck now with multi-cell?" | |
| 20:57:26 | dansmith | melwitt: better? | |
| 20:57:39 | dansmith | mriedem: s/now.*// | |
| 20:57:41 | mriedem | that's not as discoverable though | |
| 20:57:42 | mriedem | heh | |
| 20:57:49 | mriedem | especially for an end user | |
| 20:57:55 | openstackgerrit | Dan Smith proposed openstack/nova master: Add a caveat section about cellsv2 upcalls https://review.openstack.org/490612 | |
| 20:58:04 | mriedem | the end user shouldn't have any idea if the reason things don't work is because of the multi-cell decision in the cloud | |
| 20:58:12 | dansmith | well, | |
| 20:58:23 | dansmith | nor should the user be surprised that the cloud op doesn't have that filter enabled | |
| 20:58:25 | dansmith | but alas | |
| 20:59:24 | mriedem | you've got a bug in there | |
| 20:59:36 | mriedem | i'm intimately familiar with that option now | |
| 20:59:50 | dansmith | I've got a bug in who now? | |
| 20:59:55 | mriedem | your change | |
| 20:59:55 | melwitt | dansmith: yes, much better, thanks. that makes it clear IMHO | |
| 21:00:08 | mriedem | meeting starting | |
| 21:00:14 | melwitt | what's wrong with it? | |
| 21:00:35 | dansmith | oh god ugly | |
| 21:00:41 | mriedem | ha | |
| 21:01:20 | openstackgerrit | Dan Smith proposed openstack/nova master: Add a caveat section about cellsv2 upcalls https://review.openstack.org/490612 | |
| 21:09:33 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Enable cold migration with target host(2/2) https://review.openstack.org/408964 | |
| 21:09:56 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: api-ref: Add parameters in cold migrate action https://review.openstack.org/410042 | |
| 21:12:53 | dansmith | edleafe: can you push a rev to that spec to address jaypipes' typo nits at least? | |
| 21:13:24 | edleafe | dansmith: sure thing | |
| 21:13:36 | dansmith | thanks | |
| 21:18:29 | efried | jangutter Some colleagues and I are talking about revamping PCI in general and SR-IOV specifically. Please add me to reviews for blueprints and the like. | |
| 21:19:02 | jangutter | efried: will do, I should also be at the Queens PTG. | |
| 21:19:27 | efried | jangutter See you there. We'll hopefully be on the agenda with a (draft, at least) blueprint. | |
| 21:21:00 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add a caveat section about cellsv2 upcalls https://review.openstack.org/490612 | |
| 21:21:41 | dansmith | mriedem: thanks for that.. was going too fast | |
| 21:22:46 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: List/show all server migration types (1/2) https://review.openstack.org/430608 | |
| 21:26:36 | alkhodos_ | Hi everyone. I have a question about live migration. Is there an option to migrate a VM with storage? Any help or maybe a doc to look at is is much appreciated, thanks! | |
| 21:28:14 | cfriesen | dansmith: how does the "track_instance_changes" thing work? I see the call to _init_instance_info(), is there a further "track ongoing changes" that isn't covered by a "if self.track_instance_changes" test? | |
| 21:28:27 | mriedem | cfriesen: it's in the scheduler | |
| 21:28:32 | dansmith | cfriesen: when you find out, let me know | |
| 21:28:38 | mriedem | cfriesen: compute rpc casts things to the scheduler | |
| 21:28:46 | mriedem | which then updates the list of tracked instances per HostState object, | |
| 21:28:53 | mriedem | which is used by the affinity filters | |
| 21:29:13 | cfriesen | mriedem: is that part unconditional? | |
| 21:29:20 | mriedem | cfriesen: mind you, that's all computes in the deployment rpc casting to the scheduler (which is probably a single one to avoid collisions) once every minute by default | |
| 21:29:28 | mriedem | cfriesen: which part? | |
| 21:29:47 | mriedem | if you set CONF.filter_scheduler.track_instance_changes=False in nova.conf on the compute, it won't rpc cast to the scheduler | |