| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-27 | |||
| 13:57:39 | figleaf | looks like johnthetubaguy has taken care of it | |
| 13:58:16 | cdent | yes | |
| 13:58:21 | figleaf | and I have a merge conflict on the last of the RP series https://review.openstack.org/#/c/414769/ | |
| 13:58:55 | cdent | the debugging stuff that's right around bauzas patch is still in there for the current gate run, but if it happys will be cleared out | |
| 13:59:22 | figleaf | cool | |
| 13:59:44 | cdent | we're not trying to get nested rps in ocata, or are we? | |
| 14:00:08 | figleaf | cdent: yes, but they won't be doing very much at all | |
| 14:01:05 | cdent | then I guess my rp update 9 email is wrong, could you extend it? | |
| 14:02:20 | jroll | cdent: figleaf: safe to assume we didn't get the extra_specs=CUSTOM_FOO_RESOURCE_CLASS=1 thing done, right? | |
| 14:02:24 | johnthetubaguy | the ironic custom stuff, I guess we don't query that either yet, but if its in this cycle we can query it next cycle? | |
| 14:02:42 | johnthetubaguy | is looking at: https://review.openstack.org/#/c/404472 | |
| 14:02:47 | jroll | johnthetubaguy: I think we have similar questions :) | |
| 14:02:56 | cdent | jroll: the handling of extra specs, no, but the recording of inventory is still trying to get in | |
| 14:03:06 | johnthetubaguy | jroll: heh, yup | |
| 14:03:20 | figleaf | cdent: dammit, you type too fast | |
| 14:03:48 | jroll | cdent: right, so operators cannot change their flavors until pike, so we can't cut over how we schedule until q :( | |
| 14:04:27 | jroll | with 'cut over' ~= 'drop cpu/ram/disk-based scheduling' | |
| 14:04:36 | figleaf | jroll: wait - why can't they update their flavors? The new stuff will be ignored, but still... | |
| 14:04:48 | cdent | jroll: I guess so? :( It's been really hard to track some of this | |
| 14:05:17 | jroll | figleaf: because a flavor with cpu/ram/disk won't work in the new world, where ironic resource providers don't have cpu/ram/disk | |
| 14:05:40 | jroll | it would require CUSTOM_FOO *and* 2 vcpu and 16384 rams, etc | |
| 14:06:09 | johnthetubaguy | so I am reading through, it looks like each nova-compute node can only look at one ironic resource_class, is that correct? | |
| 14:06:16 | jroll | the old flavors are incompatible with the new | |
| 14:06:44 | figleaf | jroll: ah, so they could update their flavors, but the flavors will have both the resource class name *and* the vm-ish stuff too | |
| 14:07:09 | figleaf | johnthetubaguy: that's the idea | |
| 14:07:40 | jroll | figleaf: yeah, something like that | |
| 14:08:03 | johnthetubaguy | jroll: are we OK with nova-compute to resource class 1:1 mapping? | |
| 14:08:05 | figleaf | jroll: the scheduler/placement will just ignore the former for now, and at some point, switch to ignoring the latter | |
| 14:08:53 | figleaf | johnthetubaguy: to be clear, there can be multiple nova-computes that are mapped to the same resource class | |
| 14:09:37 | jroll | figleaf: I'm skeptical, how would it know to ignore ram/cpu for that flavor (but not for a virt flavor)? | |
| 14:10:09 | jroll | johnthetubaguy: that seems... wrong. are we talking nova-compute service or compute_node entry? a compute service should handle more than one resource class fine | |
| 14:10:51 | jroll | there should be n:1 node:resource_class, and m:n service:resource_class | |
| 14:11:21 | johnthetubaguy | jroll: that was my assumption | |
| 14:12:06 | figleaf | jroll: there will probably be an "if _is_ironic():" check added to that code to look for the resource class | |
| 14:12:12 | jroll | johnthetubaguy: I also don't see that n:1 service:resource_class in the code | |
| 14:12:25 | johnthetubaguy | jroll: the resource class request's presence just always dominates any other request, that seems fine | |
| 14:13:57 | cdent | Is it normal for us to be as confused about stuff as we've been? | |
| 14:14:04 | johnthetubaguy | figleaf: I guess we can make it nested, once we have nested? | |
| 14:14:43 | johnthetubaguy | cdent: thats probably yes, I prefer the "can we do better at this" question, I hope thats a yes too | |
| 14:15:00 | cdent | just checking | |
| 14:15:04 | johnthetubaguy | jroll: I am looking here: https://review.openstack.org/#/c/404472/23/nova/compute/resource_tracker.py@525 | |
| 14:15:11 | johnthetubaguy | cdent: its good to check these things | |
| 14:15:18 | cdent | I'm trying to decide whether it is reasonable to be chill or panicky | |
| 14:15:57 | jroll | johnthetubaguy: yeah, so that's for a single ironic node, and that code will run once per ironic node | |
| 14:16:29 | openstackgerrit | Andrey Volkov proposed openstack/nova: WIP Update pci device fields from deleted object https://review.openstack.org/426243 | |
| 14:16:46 | figleaf | cdent: it's more the overlapping terminology that confuses me | |
| 14:16:56 | johnthetubaguy | jroll: doh, I see the loop now | |
| 14:17:05 | figleaf | "compute" can mean multiple things | |
| 14:18:00 | johnthetubaguy | figleaf: sorry, I see what you were trying to tell me now | |
| 14:18:08 | jroll | figleaf: I've learned to always say compute *service* (nova-compute process) or compute *node* (the hypervisor/hardware), it seems to help | |
| 14:18:33 | cdent | jroll: if you said compute service to me, I'd assume you meant all of nova | |
| 14:19:00 | jroll | cdent: ahem, "nova-compute service" | |
| 14:19:00 | cdent | language is hard | |
| 14:19:02 | jroll | heh | |
| 14:21:39 | openstackgerrit | Sam Betts proposed openstack/nova: Change order of _cleanup_deploy and _unprovision in Ironic virt https://review.openstack.org/422678 | |
| 14:21:40 | jroll | through an IV | |
| 14:23:48 | cdent | jroll++ | |
| 14:26:30 | artom | Intra-vineous :D | |
| 14:26:38 | figleaf | johnthetubaguy: in that case, let's make some wine! | |
| 14:28:24 | openstackgerrit | Andrey Volkov proposed openstack/nova: WIP Update pci device fields from deleted object https://review.openstack.org/426243 | |
| 14:36:49 | johnthetubaguy | figleaf: best idea I heard all day | |
| 14:39:19 | BlackDex | hello again. I'm having trouble with nova-consoleauth | |
| 14:39:31 | BlackDex | i have an ha setup, with hacluster and a vip | |
| 14:39:37 | BlackDex | 3 systems | |
| 14:40:01 | BlackDex | if i use horizon to view the console i have a 1 in 3 times luck to get the console to view | |
| 14:40:42 | BlackDex | i have configured memcached, but nothing is getting into memcached, i can connect to memcached from the nova-controle nodes | |
| 14:47:20 | superdan | that scheduler change failed on the old side due to volumes and neutron | |
| 14:47:21 | cdent | johnthetubaguy, dansmith : looks like the latest multinode grenade failed on timeouts this time | |
| 14:47:25 | cdent | jinx | |
| 14:47:27 | superdan | heh | |
| 14:48:16 | johnthetubaguy | is that a real something is broken timeout, or a you hit an unlucky node timeout? | |
| 14:49:18 | johnthetubaguy | cdent: did we get the correct min_version back now? | |
| 14:49:28 | cdent | no logs for the new side | |
| 14:49:53 | superdan | johnthetubaguy: we didn't get far enough to look | |
| 14:50:16 | johnthetubaguy | I was worried you would say that | |
| 14:50:36 | johnthetubaguy | heh | |
| 14:50:43 | mriedem | superdan: congrats https://review.openstack.org/319379 | |
| 14:51:17 | mriedem | now we just cleanup the release notes for cells v2 and maybe upgrade docs and we're good for ocata i think | |
| 14:51:35 | jroll | what's the best way to talk to the placement service in devstack? just grab a keystone token and use curl? | |
| 14:51:50 | mriedem | jroll: yeah, althouh mlavalle has an osc plugin somewhere | |
| 14:52:17 | jroll | mriedem: cool, thanks | |
| 14:52:19 | cdent | jroll: a) probably yes b) i'm not sure if mlavalle's thing is "done" c) I use gabbi yaml files, run throuh gabbi-run | |
| 14:52:37 | johnthetubaguy | figleaf: cdent: I am struggling to understand this line: https://review.openstack.org/#/c/404472/23/nova/compute/resource_tracker.py@548 | |
| 14:52:41 | openstackgerrit | Artom Lifshitz proposed openstack/nova: Fix tag attribute disappearing in 2.33 and 2.37 https://review.openstack.org/424759 | |
| 14:52:41 | openstackgerrit | Artom Lifshitz proposed openstack/nova: Pass APIVersionRequest to extensions https://review.openstack.org/425876 | |
| 14:53:05 | johnthetubaguy | figleaf: cdent: I feel I am missing something big and obvious | |
| 14:53:10 | jroll | cdent: thanks | |
| 14:53:11 | cdent | jroll: in case you're curious: https://gist.github.com/cdent/a9590764fbc7402d450fa36df14f35e0 | |
| 14:53:42 | jroll | cdent: I need to poll for resources in our devstack plugin, curl seems like the right thing for now | |
| 14:53:52 | cdent | jroll: for that, yeah | |
| 14:54:18 | cdent | (although, just because we're almost on the topic, gabbi can poll) | |
| 14:54:48 | cdent | johnthetubaguy: the cleanup or that the exception happens? | |
| 14:55:14 | johnthetubaguy | cdent: I am wondering if we need cleanup in other cases, like when an ironic node is there, but there is no allocation | |
| 14:55:26 | johnthetubaguy | cdent: maybe the API just does that for us already? | |
| 14:55:56 | johnthetubaguy | ah, I think it does | |
| 14:56:04 | cdent | I don't think I'm sufficiently contextualized to really follow what you're saying? | |
| 14:56:43 | openstackgerrit | Artom Lifshitz proposed openstack/nova-specs: Fix tag attribute disappearing https://review.openstack.org/426030 | |
| 14:57:08 | johnthetubaguy | cdent: sorry, I was worried about us not always deleting the old style ironic inventories, but I didn't realise the API would do that for us when we put the inventories | |
| 14:57:39 | mriedem | superdan: cdent: jaypipes: bauzas: johnthetubaguy: melwitt: fyi i'm starting to make a list of things to do before rc1 https://etherpad.openstack.org/p/nova-ocata-rc1-todos | |
| 14:57:44 | johnthetubaguy | its just when we have allocations we need to do better | |