| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-09 | |||
| 10:01:48 | bauzas | jianghuaw_: around ? | |
| 10:01:56 | bauzas | jianghuaw_: just a clarification with https://review.openstack.org/#/c/516899/34/nova/virt/xenapi/driver.py | |
| 10:02:44 | bauzas | jianghuaw_: since you don't lookup the allocations (just verifying whether there are some VGPU ones), you don't verify how many RPs allocations are there, and just say "okay, only one", right? | |
| 10:03:45 | bauzas | jianghuaw_: for example, say I have an allocations list having two allocations for two separate RPs, and each one having VGPU:2, then you just ask Xen to create only one (not 4), right? | |
| 10:06:41 | jianghuaw_ | Bauzas, yes xen only supports one vGPU per VM. | |
| 10:06:52 | bauzas | that, I know | |
| 10:06:58 | bauzas | I was just wanting to make sure | |
| 10:07:10 | jianghuaw_ | bauzas, why it will pass down multiple allocations? | |
| 10:07:23 | bauzas | jianghuaw_: it won't | |
| 10:07:47 | bauzas | jianghuaw_: just the fact that you'll need to modify your method once we use nested RPs | |
| 10:08:21 | bauzas | but at least, you could have one allocation for a RP, with a value more than 1, but okay | |
| 10:08:53 | bauzas | jianghuaw_: FWIW, the nvidia driver for libvirt doesn't accept more than one vGPU per pGPU | |
| 10:09:01 | openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Use neutron port_list when filtering instance by ip https://review.openstack.org/525505 | |
| 10:09:16 | jianghuaw_ | bauzas, ah, Thanks. we need change that method. I think naichuan already has one patch to mapping the RP-uuid from the allocation to the proper GPU group. | |
| 10:09:35 | bauzas | jianghuaw_: but you can ask for 2 vGPUs if you have more than 1 pGPU for your host | |
| 10:10:09 | jianghuaw_ | no. still only one vGPU. | |
| 10:10:26 | bauzas | jianghuaw_: WDYM ? | |
| 10:10:28 | jianghuaw_ | It don't support >1 vGPU for one VM. | |
| 10:10:35 | bauzas | for nvidia ? | |
| 10:11:10 | bauzas | AFAIK, you can have 2 vGPUs with the nvidia driver *if* each of them are by a separate physical card | |
| 10:11:13 | jianghuaw_ | XenServer has the restriction in xapi. | |
| 10:11:32 | bauzas | oh ok, I misunderstood | |
| 10:11:43 | jianghuaw_ | I guess nvidia can support but xen can't. | |
| 10:11:52 | bauzas | no, nvidia driver can't | |
| 10:12:05 | bauzas | jianghuaw_: http://images.nvidia.com/content/grid/pdf/GRID-vGPU-User-Guide.pdf section 1.3.2 | |
| 10:12:42 | bauzas | it tells to use homogenous vGPU types, but it also means that you need to use separate physical devices, at least for libvirt | |
| 10:13:21 | jianghuaw_ | can it support multiple vGPU with the same vGPUtype? | |
| 10:14:04 | bauzas | like I said, you need to use separate physical devices | |
| 10:14:17 | bauzas | that's a driver limitation | |
| 10:15:52 | jianghuaw_ | bauzas, Got it. thanks for letting me know of it. | |
| 10:16:42 | jianghuaw_ | bauzas, do you have a link on the *driver limitation*? | |
| 10:16:50 | bauzas | no, unfortunately | |
| 10:17:05 | bauzas | that's something someone told me | |
| 10:17:16 | bauzas | awesome, heh ? | |
| 10:17:37 | bauzas | code driven by trust | |
| 10:17:49 | bauzas | and word of mouth | |
| 10:18:09 | jianghuaw_ | :-) | |
| 10:20:17 | openstackgerrit | Peter Penchev proposed openstack/nova master: Add the Nova libvirt StorPool attachment driver. https://review.openstack.org/140733 | |
| 10:35:53 | openstackgerrit | Stephen Finucane proposed openstack/nova master: console: introduce the VeNCrypt RFB authentication scheme https://review.openstack.org/345398 | |
| 10:35:54 | openstackgerrit | Stephen Finucane proposed openstack/nova master: console: Provide an RFB security proxy implementation https://review.openstack.org/345399 | |
| 10:35:54 | openstackgerrit | Stephen Finucane proposed openstack/nova master: doc: Document TLS security setup for noVNC proxy https://review.openstack.org/500544 | |
| 10:35:55 | openstackgerrit | Stephen Finucane proposed openstack/nova master: console: Send bytes to sockets https://review.openstack.org/531834 | |
| 10:39:01 | openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Use neutron port_list when filtering instance by ip https://review.openstack.org/525505 | |
| 10:47:37 | harsha | Hi folks, quick question on some python unit tests; I need to run - 'tox -e pep8'; should i run this on the test machine in the directory/path where I have my changes ? | |
| 10:47:51 | tetsuro | stephenfin: thanks for reviewing https://review.openstack.org/#/c/531328/ :) I'm both okay with microversion or not. | |
| 10:48:50 | tetsuro | stephenfin: Just FYI, I happend to find this blog https://techglimpse.com/why-does-openstack-report-hypervisor-type-as-qemu-when-libvirt_type-is-kvm/ . I don't know who wrote this, though. | |
| 10:49:27 | stephenfin | tetsuro: Oh yeah, to be fair, I honestly don't know what way it should be hence no -1 | |
| 10:49:44 | stephenfin | cdent: Fancy weighing in on https://review.openstack.org/#/c/531328/ ? | |
| 10:50:09 | stephenfin | I know we fix some bugs without microversions, but maybe they're just for 5xx type errors | |
| 10:50:14 | sean-k-mooney | tetsuro: well technicall kvm is not a hypervisor | |
| 10:50:32 | sean-k-mooney | it is kernel acclartion module that is used to acclerate qemu | |
| 10:51:19 | tetsuro | sean-k-mooney: hi, happy new year! | |
| 10:51:49 | sean-k-mooney | tetsuro: same to you :) i trust you had a good vaction | |
| 10:52:59 | tetsuro | sean-k-mooney: yup, I had a lot of mochi(rice cake) in new year days. | |
| 10:53:27 | cdent | stephenfin: looking | |
| 10:54:39 | rabel_b1 | do you know, what happened to https://docs.openstack.org/user-guide/cli-nova-launch-instance-from-volume.html ? i cannot seem to find it anywhere in the nova docs | |
| 10:54:55 | tetsuro | It is true that KVM is a kind of acceralator for QEMU, but in OpenStack document, it is one of the hypervisors. https://docs.openstack.org/mitaka/config-reference/compute/hypervisors.html | |
| 10:55:26 | lyarwood | mdbooth: https://review.openstack.org/#/c/460243 btw, merged both, pretty large change to backport but meh. | |
| 10:57:30 | hrw | tetsuro: kvm is in-kernel hypervisor with qemu used as interface to it. sth like that iirc | |
| 10:58:06 | mdbooth | lyarwood: Doesn't look that bad | |
| 10:58:11 | mdbooth | Not in driver.py, anyway | |
| 11:06:59 | cdent | you're probalby right | |
| 11:08:15 | sean-k-mooney | cdent: well when it comes to micro version is not really someting i spend much time thinking about normally so im not sure about that but im leaning towords the bump in this instance | |
| 11:08:42 | cdent | you've provided the critical point: a client may make decision based on the value | |
| 11:30:37 | openstackgerrit | Merged openstack/nova master: Updated from global requirements https://review.openstack.org/531737 | |
| 11:34:36 | kashyap | tetsuro: KVM runs part of the guest instructions directly on the baremetal CPU (taking advantage of CPU Virtualization extensions - Intel-VMX / AMD-SVM). While QEMU provides rest of the PC-like components (n/w card, disk interface, etc) | |
| 11:36:55 | openstackgerrit | Jie Li proposed openstack/nova master: Support volume-backed server rescue https://review.openstack.org/531524 | |
| 11:37:11 | tetsuro | hrw, kashyap: good to know. thanks for the information. | |
| 11:38:09 | tetsuro | sean-k-mooney, cdent: thank you for reviewing, let me wait for matt to leave comments here. | |
| 11:39:03 | sandanar | hi.. does anyone know what is the reason behind CONF.libvirt.volume_use_path always returns the default value and not the one passed in [libvirt] section of /etc/nova/nova.conf - https://bugs.launchpad.net/nova/+bug/1723928 this is causing multipath failures on the cinder side | |
| 11:39:04 | openstack | Launchpad bug 1723928 in os-brick "In case of volume_use_multipath=True, Nova unable to fetch CONF.libvirt.volume_use_multipath value from nova.conf" [Undecided,New] | |
| 11:40:45 | openstackgerrit | Yikun Jiang (Kero) proposed openstack/python-novaclient master: Microversion 2.59 - Migrations list pagination https://review.openstack.org/516168 | |
| 11:41:17 | lyarwood | kashyap: http://paste.openstack.org/show/641098/ - just noticed this on stable/pike https://review.openstack.org/#/c/531407/ | |
| 11:41:38 | lyarwood | kashyap: load of migration failed: Operation not permitted | |
| 11:42:40 | kashyap | lyarwood: That rings a bell, I recall seeing a bug for the VQ size | |
| 11:48:00 | kwathore | @meriedm: This is very critical bug which we have observerd releated to multipath: https://bugs.launchpad.net/nova/+bug/1723928 | |
| 11:48:01 | openstack | Launchpad bug 1723928 in os-brick "In case of volume_use_multipath=True, Nova unable to fetch CONF.libvirt.volume_use_multipath value from nova.conf" [Undecided,New] | |
| 11:50:30 | hrw | hm. https://review.openstack.org/#/c/530965/ got rebased 2h ago and zuul still did not started | |
| 11:55:02 | openstackgerrit | Merged openstack/nova stable/newton: Raise MarkerNotFound if BuildRequestList.get_by_filters doesn't find marker https://review.openstack.org/530982 | |
| 11:55:34 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Add support for Windows network commands https://review.openstack.org/487405 | |
| 11:56:29 | kashyap | lyarwood: From the top off my head, I vaguely recall a couple of fixes from Virtio net devices from StefanH. The bug links are eluding me | |
| 11:56:48 | kashyap | I asked Stefan for the specific versions that fix these screwey virtio queue issues | |
| 11:57:01 | kashyap | Will let ya know once I hear back from him | |
| 11:57:04 | lyarwood | kashyap: cool thanks | |
| 11:59:39 | openstackgerrit | Merged openstack/nova master: Add regression test for resize failing during retries https://review.openstack.org/531405 | |
| 12:00:53 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Add memoize function using oslo.cache https://review.openstack.org/472773 | |
| 12:03:21 | hrw | https://review.openstack.org/#/c/530965/ finally got +1 from Zuul. now just have to get second run of Zuul for +2 and merge ;d | |
| 12:03:50 | openstackgerrit | sahid proposed openstack/nova master: hardware: only take into account small pages https://review.openstack.org/532168 | |
| 12:16:48 | openstackgerrit | Yikun Jiang (Kero) proposed openstack/python-novaclient master: Microversion 2.59 - Migrations list pagination https://review.openstack.org/516168 | |
| 12:19:44 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Add abstract OVSDB API https://review.openstack.org/476612 | |
| 12:27:13 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Add native implementation OVSDB API https://review.openstack.org/482226 | |
| 12:28:56 | ildikov | johnthetubaguy: gibi: hi :) this tiny patch needs a second +2 if either of you have a few minutes: https://review.openstack.org/#/c/529695/ | |
| 12:29:10 | ildikov | johnthetubaguy: gibi: thanks in advance :) | |
| 12:38:59 | openstackgerrit | Lajos Katona proposed openstack/nova master: Deduplicate aggregate notification samples https://review.openstack.org/531162 | |
| 12:39:36 | gibi | ildikov: done :) | |
| 12:44:32 | openstackgerrit | Lajos Katona proposed openstack/nova master: Deduplicate service status notification samples https://review.openstack.org/531381 | |
| 12:44:48 | ildikov | gibi: thanks!! :) | |
| 13:06:33 | sean-k-mooney | stephenfin: just looking at https://review.openstack.org/#/c/531358 looks like when you change your email to stephenfin instead of sfinucan you didnt update your global git user name. not sure if thats intentional but just taught i would point it out | |
| 13:06:50 | sean-k-mooney | * global git email | |
| 13:08:12 | openstackgerrit | Lajos Katona proposed openstack/nova master: Deduplicate aggregate notification samples https://review.openstack.org/531162 | |