| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-02-04 | |||
| 12:46:26 | openstackgerrit | Stephen Finucane proposed openstack/nova master: mypy: Add type annotations to 'nova.pci' https://review.opendev.org/676209 | |
| 12:46:27 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Bump minimum version of websockify https://review.opendev.org/705654 | |
| 12:46:27 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Address TODO https://review.opendev.org/705653 | |
| 12:46:28 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Remove 'run_once' helper https://review.opendev.org/705656 | |
| 12:46:28 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Merge unnecessary 'NovaProxyRequestHandlerBase' separation https://review.opendev.org/705655 | |
| 12:46:29 | openstackgerrit | Stephen Finucane proposed openstack/nova master: WIP: mypy: Add type annotations to top-level modules https://review.opendev.org/705658 | |
| 12:46:29 | openstackgerrit | Stephen Finucane proposed openstack/nova master: mypy: Add nova.cmd, nova.conf, nova.console https://review.opendev.org/705657 | |
| 12:53:47 | openstackgerrit | Stephen Finucane proposed openstack/nova master: nova-net: Remove unused parameters https://review.opendev.org/703974 | |
| 12:58:15 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Address release note nits for cpu-resources series https://review.opendev.org/705667 | |
| 12:58:15 | openstackgerrit | Stephen Finucane proposed openstack/nova master: hardware: Add TODO to remove '(un)pin_cpu_with_siblings' https://review.opendev.org/705666 | |
| 12:58:16 | openstackgerrit | Stephen Finucane proposed openstack/nova master: doc: Address some trivial nits with port QoS doc https://review.opendev.org/705668 | |
| 12:58:43 | openstackgerrit | Merged openstack/nova master: Enable live migration with qos ports https://review.opendev.org/699066 | |
| 12:58:48 | openstackgerrit | Merged openstack/nova master: Avoid fetching metadata when no subnets found https://review.opendev.org/679247 | |
| 13:02:23 | openstackgerrit | Chen Hanxiao proposed openstack/nova master: libvirt: don't call sync_guest_time if qga is not enabled https://review.opendev.org/524836 | |
| 13:30:24 | huaqiang | stephenfin: do you have updated comment for the mixed instance spec https://review.opendev.org/#/c/668656/ ? | |
| 13:39:06 | kashyap | sfinucan: or any rST-aware-human: On headings automatically getting anchors -- is that a Sphinx thing, or rST proper? | |
| 14:09:26 | sean-k-mooney | kashyap: it depens on your theme i think | |
| 14:10:16 | kashyap | The theme is Sphinx in this case (as alluded to above) | |
| 14:10:23 | sean-k-mooney | kashyap: but heading in lists dont get css ids allcoated | |
| 14:11:41 | sean-k-mooney | so if you look at the flavor docs for example because we are using a list none of the falvor exrta specs have clickable ancors | |
| 14:12:42 | sean-k-mooney | the title in the list will stilll be bolded but the element that renders the list does not add the css id | |
| 14:13:24 | sean-k-mooney | kashyap: i am assuming you are asking why overview has an ancor bug flavor ID does not right https://docs.openstack.org/nova/latest/user/flavors.html#overview | |
| 14:14:25 | kashyap | sean-k-mooney: No, not really; the example is: | |
| 14:19:47 | kashyap | sean-k-mooney: Compare and contrast: http://paste.openstack.org/show/789109/ | |
| 14:20:09 | kashyap | The bottom example is the round-about way; the first is the straight-forward way. Anyway | |
| 14:20:30 | sean-k-mooney | oh you are tralking about the automatic refrences | |
| 14:20:48 | sean-k-mooney | so that you can refer to it without needing to do .. _`first-section`: | |
| 14:21:04 | kashyap | Yep | |
| 14:21:13 | sean-k-mooney | ya that only works in the current doc/page | |
| 14:21:27 | sean-k-mooney | if you need to reference someitn in a different rst file you have to do it manually | |
| 14:22:09 | sean-k-mooney | `First Section`__ will not work aross rst files | |
| 14:25:42 | kashyap | Right, just within the file :-) All sorted | |
| 14:26:15 | sean-k-mooney | yep we only use the _`first-section`: way when we need too | |
| 14:26:22 | kashyap | sean-k-mooney: Unrelated, do you know how to look up what version of Debian does this package edk2-(0~20190606.20d2e5a1-2) have? | |
| 14:26:33 | bauzas | efried: sean-k-mooney: had fully read the etherpad and the proposals | |
| 14:26:36 | bauzas | <3 for you | |
| 14:26:46 | bauzas | now processing those | |
| 14:26:49 | kashyap | Rephrasing: "... what version of Debian has this package edk2-(0~20190606.20d2e5a1-2)" | |
| 14:28:21 | sean-k-mooney | bauzas: cool, did it seam familar? i have proposed it in the past but with same same_subtree and a few other enhancements to placmeet it is now much easier to use | |
| 14:28:46 | bauzas | sean-k-mooney: same_subtree should be good, yes | |
| 14:28:58 | bauzas | sean-k-mooney: efried: just a question about upgrades | |
| 14:29:28 | bauzas | I provided a config option for listing the resource types | |
| 14:29:45 | bauzas | I understand you wouldn't want to use it | |
| 14:30:15 | bauzas | but then that means that when we upgrade, then we would have to transform the inventories and allocations directly | |
| 14:30:19 | bauzas | I'm cool with this | |
| 14:30:25 | bauzas | but you all okay? | |
| 14:31:23 | sean-k-mooney | i was ok with the config option | |
| 14:31:49 | sean-k-mooney | we will still need at least a config option to say report numa or not. where it needs to be a list is debatable | |
| 14:32:06 | sean-k-mooney | i was ok with the list | |
| 14:32:06 | openstack | Launchpad bug 1861876 in OpenStack Compute (nova) "[Neutron API] Neutron Floating IP not always have 'port_details'" [Undecided,New] | |
| 14:32:06 | ralonsoh | sean-k-mooney, stephenfin https://bugs.launchpad.net/nova/+bug/1861876 | |
| 14:33:12 | sean-k-mooney | ralonsoh: looking | |
| 14:33:16 | ralonsoh | thanks | |
| 14:33:41 | sean-k-mooney | oh the nova net removal | |
| 14:33:46 | ralonsoh | yeah! | |
| 14:33:52 | ralonsoh | just a couple of things | |
| 14:34:21 | openstackgerrit | Mykola Yakovliev proposed openstack/nova master: Fix boot_roles in InstanceSystemMetadata https://review.opendev.org/698040 | |
| 14:34:31 | sean-k-mooney | ralonsoh: is this the issue https://review.opendev.org/#/c/697153/16/nova/api/openstack/compute/floating_ips.py@39 | |
| 14:34:52 | ralonsoh | sean-k-mooney, exactly | |
| 14:35:02 | sean-k-mooney | we shoudl be useing floating_ip.get('prot_detail') | |
| 14:35:10 | sean-k-mooney | to not cause a key error if it is not set | |
| 14:35:13 | ralonsoh | sean-k-mooney, and the way the "pool" key is retrieved | |
| 14:35:29 | ralonsoh | the network id is stored in "floating_network_id" | |
| 14:35:37 | ralonsoh | https://github.com/openstack/neutron/blob/master/neutron/db/l3_db.py#L1030-L1037 | |
| 14:35:38 | sean-k-mooney | ah i see | |
| 14:36:03 | sean-k-mooney | im glad we are so consistent | |
| 14:36:18 | ralonsoh | hahahahaha | |
| 14:36:21 | ralonsoh | sorry for that | |
| 14:36:34 | sean-k-mooney | no its fine so that shoudl be an easy fix | |
| 14:36:52 | efried | bauzas, sean-k-mooney: I really don't want the list. I think a boolean toggle is the right thing. As we move forward, for example moving VGPUs from under the root to under the NUMA RPs, things will still work correctly. | |
| 14:37:28 | sean-k-mooney | efried: ya im totally fine with the boolean toggel | |
| 14:37:37 | bauzas | efried: we don't really need a boolean for triggering if so | |
| 14:37:38 | efried | We'll "gain" support for new flavors that express affinity of the VGPUs, but old flavors that don't express such affinity will still work. | |
| 14:37:59 | efried | bauzas: we do; one of the important distinctions here is that we're segregating the data center into numa-aware and not. | |
| 14:38:00 | sean-k-mooney | i think by defualt we shoudl not do the reshape on upgrade but when the config option is set we will do the reshape and it should not be possibel to undo while instance are on the host | |
| 14:38:03 | bauzas | unless we wanna somehow prepare ops to switch when they want | |
| 14:38:32 | bauzas | cool enough, let's then make it a boolean | |
| 14:38:46 | sean-k-mooney | bauzas: we need the config option because we said we would partion the cloud into numa hosts and non numa hosts | |
| 14:38:50 | efried | I agree we should reshape when the config option is set and the service is restarted. We knew the restriction of "reshape only on upgrade" was artificial/temporary when we made it. | |
| 14:39:12 | sean-k-mooney | so that on the non numa hosts you could continue to run floating instance that use more resouces then fit in 1 numa node | |
| 14:39:20 | sean-k-mooney | without haveing to make the multi numa instnces | |
| 14:39:43 | bauzas | that looks seamless indeed | |
| 14:39:43 | stephenfin | ralonsoh: ah, so previously we were making two requests to neutron via neutronclient: one to retrieve all floating IPs and one to retrieve all ports | |
| 14:39:57 | bauzas | but my only concern is that this conf opt is only for Ussuri | |
| 14:40:00 | sean-k-mooney | e.g. the i want to use openstack to run one giant vm per host to run anohter orcastrator on top usecase we all hate | |
| 14:40:11 | bauzas | and then we would turn into it either way in Victoria | |
| 14:40:17 | sean-k-mooney | bauzas: i wont be | |
| 14:40:20 | dansmith | efried: sean-k-mooney bauzas but they will have to reshape eventually, right? | |
| 14:40:22 | stephenfin | ralonsoh: and then we simply matched 'port_id' for each entry in the former to the corresponding port from the latter. I tried to make that cleverer by using 'port_details', but you're saying that's an extension that I can't rely on | |
| 14:40:29 | sean-k-mooney | i will need to be kept as long as we support non numa instnaces | |
| 14:40:30 | bauzas | dansmith is expressing my concern | |
| 14:40:39 | stephenfin | ralonsoh: I can fix that now if you haven't already, but how can I check if that extension is available? | |
| 14:40:41 | efried | dansmith: I'm not sure we ever need to force them to make a host NUMA-aware if they don't want to. | |
| 14:40:56 | ralonsoh | stephenfin, exactly, this is an extension and this key is not mandatory | |
| 14:41:12 | bauzas | efried: if so, we somehow need to make the path clear that we *will* remove filter things in Victoria either way | |
| 14:41:14 | sean-k-mooney | dansmith: they will have to reshape only if we decied that all instance will have a numa toplogy of 1 gust numa node by defualt | |
| 14:41:18 | efried | iow the segregated-on-NUMA-ness is a permanent characteristic of the data center | |
| 14:41:58 | dansmith | if they can stay off forever, then fine, but I imagine that leaves us with a pretty big variable for a long time | |
| 14:42:05 | bauzas | also, | |
| 14:42:23 | bauzas | what I'm not okay is keeping the old filter processing for NUMA placement for a while | |