| gmann | tc-members: if you would like to add something more for community meeting please do, I will start finalizing it on Thursday morning - https://etherpad.opendev.org/p/openstack-updates-community-meeting | 00:18 |
|---|---|---|
| *** tetsuro has joined #openstack-tc | 00:42 | |
| diablo_rojo | o/ | 01:01 |
| gmann | o/ | 01:01 |
| zaneb | o/ | 01:01 |
| zaneb | diablo_rojo: I was thinking today that maybe we should define a starter-kit:virtualised-k8s tag | 01:28 |
| zaneb | it would be starter-kit:compute + Manila, Designate, Octavia, Barbican | 01:28 |
| zaneb | wdyt? | 01:28 |
| fungi | constellations ii: the return | 01:30 |
| fungi | in all seriousness though, i concur, more starter-kit:foo formations or similar would be beneficial, and that seems like an obvious next choice | 01:32 |
| zaneb | I wasn't at the meeting where it was dreamed up, so I never understood how constellations were different from the tag thing | 01:32 |
| fungi | they weren't particularly | 01:33 |
| fungi | the idea was that you have this sky full of projects, and you connect some subset of them to serve a common use case | 01:33 |
| fungi | though "sky full of projects" does harken back to "we've got rather a lot of projects" | 01:34 |
| fungi | so constellation as an analogy may have been a poor choice if we wanted to de-emphasize that particular bit of our reality | 01:34 |
| zaneb | a thousand points of light :D | 01:35 |
| fungi | and now you're harkening me back to something else i'd rather forget | 01:35 |
| zaneb | I fear we may be the only ones here old enough to recognise that reference | 01:36 |
| fungi | i'm perfectly okay with that | 01:36 |
| fungi | anyway, my take on the constellations idea was that we could tie it to the project map and make it an interactive display, so you'd select a particular use case and it would highlight the appropriate components on the map | 01:41 |
| fungi | the original bolded set was sort of a take on that, i guess, though not fully realized and more closely aligned with the openstack powered platform trademark than something like starter-kit:compute | 01:43 |
| fungi | also someone felt compelled to put the word "core" on there... almost as if they were trolling us | 01:44 |
| zaneb | yes, it's as if somebody _likes_ sowing confusion | 01:51 |
| fungi | part of me wants to blame mckenty | 01:55 |
| fungi | though he was gone by then, his spirit lived on | 01:55 |
| *** ricolin has joined #openstack-tc | 02:01 | |
| openstackgerrit | Ghanshyam Mann proposed openstack/project-team-guide master: Clarify deprecation tags and record deprecation cycle https://review.opendev.org/736082 | 02:26 |
| openstackgerrit | Ghanshyam Mann proposed openstack/governance master: Record deprecated cycle for deprecated project https://review.opendev.org/736084 | 02:31 |
| openstackgerrit | Ghanshyam Mann proposed openstack/project-team-guide master: Clarify deprecation tags and record deprecation cycle https://review.opendev.org/736082 | 02:31 |
| diablo_rojo | zaneb, I think thats a good idea. | 02:45 |
| diablo_rojo | It would be good to kind of.. easily see what services work with k8s | 02:46 |
| diablo_rojo | I guess cinder should have one too? and keystone? | 02:46 |
| diablo_rojo | Unless I am misunderstanding. | 02:46 |
| fungi | i've heard those are commonly used for block storage backends with containers, and multi-tenant kubernetes separation | 02:47 |
| fungi | also neutron+kuryr apparently? | 02:48 |
| diablo_rojo | Oh yeah that could be another. | 02:49 |
| diablo_rojo | I was just thinking about all the plugins and whatnot in the provider openstack repo in k8s | 02:50 |
| *** markvoelker has joined #openstack-tc | 04:27 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** markvoelker has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-tc | 04:33 | |
| *** diablo_rojo has quit IRC | 04:53 | |
| *** markvoelker has joined #openstack-tc | 06:09 | |
| *** markvoelker has quit IRC | 06:14 | |
| *** rpittau|afk is now known as rpittau | 06:56 | |
| *** slaweq has joined #openstack-tc | 07:00 | |
| *** dklyle has quit IRC | 07:25 | |
| *** tosky has joined #openstack-tc | 07:36 | |
| *** ralonsoh has joined #openstack-tc | 07:57 | |
| *** e0ne has joined #openstack-tc | 08:15 | |
| ttx | re: renaming master branches, I think the question is not whether the term was really meant in its master/slave sense. If our industry in general is moving away from that term, I think we should, too. | 08:48 |
| ttx | So yes, we should evaluate how we would do it | 08:50 |
| *** tkajinam has quit IRC | 08:57 | |
| *** e0ne has quit IRC | 09:02 | |
| *** e0ne has joined #openstack-tc | 09:08 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/governance master: Add openstack-ansible-os_adjutant repo https://review.opendev.org/736140 | 09:26 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/governance master: Add openstack-ansible-os_adjutant repo https://review.opendev.org/736140 | 09:31 |
| *** e0ne has quit IRC | 09:47 | |
| *** e0ne_ has joined #openstack-tc | 09:47 | |
| *** e0ne_ has quit IRC | 09:51 | |
| *** e0ne has joined #openstack-tc | 09:52 | |
| *** rpittau is now known as rpittau|bbl | 10:18 | |
| *** markvoelker has joined #openstack-tc | 10:29 | |
| *** markvoelker has quit IRC | 10:34 | |
| *** e0ne has quit IRC | 10:42 | |
| *** e0ne has joined #openstack-tc | 10:43 | |
| *** e0ne has quit IRC | 10:44 | |
| *** e0ne has joined #openstack-tc | 10:45 | |
| *** e0ne has quit IRC | 10:46 | |
| *** e0ne has joined #openstack-tc | 10:46 | |
| *** purplerbot has joined #openstack-tc | 11:00 | |
| *** tetsuro_ has joined #openstack-tc | 11:42 | |
| *** tetsuro has quit IRC | 11:46 | |
| cloudnull | o/ | 11:46 |
| *** tetsuro_ has quit IRC | 11:51 | |
| *** tosky has quit IRC | 11:56 | |
| *** tetsuro has joined #openstack-tc | 11:57 | |
| *** tosky has joined #openstack-tc | 11:59 | |
| *** rpittau|bbl is now known as rpittau | 12:06 | |
| *** ijolliffe has joined #openstack-tc | 12:28 | |
| jungleboyj | o/ | 12:47 |
| *** slaweq_ has joined #openstack-tc | 12:47 | |
| *** slaweq has quit IRC | 12:48 | |
| *** slaweq_ is now known as slaweq | 13:03 | |
| zaneb | fungi: +1 for Kuryr, forgot about that | 13:07 |
| zaneb | Neutron & Keystone are already in starter-kit:compute, so I was including those | 13:08 |
| zaneb | Cinder isn't actually really necessary if all you're doing is running k8s | 13:09 |
| openstackgerrit | Thierry Carrez proposed openstack/project-team-guide master: Update release management page https://review.opendev.org/736213 | 13:12 |
| fungi | zaneb: yeah, the connection i think is you can apparently use cinder drivers directly with kubernetes for an optional storage layer | 13:13 |
| fungi | at least i've heard that's fairly popular | 13:14 |
| fungi | i think it's done without interfacing with the cinder rest api, but i don't really know that much about it | 13:15 |
| *** tetsuro has quit IRC | 13:15 | |
| zaneb | fungi: you can, but k8s containers operate on file storage, not block storage (with one caveat). so there's a hacky conversion layer in there to basically give you what Manila gives you out of the box | 13:15 |
| fungi | aha, got it | 13:15 |
| fungi | so say you have an emc san... manila can just use that directly without needing a block storage driver of some sort? | 13:16 |
| fungi | (like, does manila speak fc-scsi and stuff?) | 13:16 |
| fungi | for provisioning/scaling the block devices its filesystems reside on, i mean | 13:17 |
| zaneb | the caveat is that (as of very recently) there is such a thing as mounting block storage in containers. but it's mostly there for KubeVirt, which you won't be running if you're running k8s in virt already | 13:17 |
| fungi | i suppose it's more about speaking the control/management protocols for your san, rather than the storage layer protocols | 13:17 |
| tbarron | so cinder-csi works well for k8s with containers running on VMs | 13:17 |
| tbarron | but it's written with the nova-cinder coupling | 13:18 |
| tbarron | so I don't think it works today for containers on bare metal compute instances | 13:18 |
| fungi | got it. thanks, that's useful detail | 13:18 |
| tbarron | Have been checking that out post-teapot fwiw | 13:19 |
| zaneb | tbarron: context here is running in virt. but did not know that, thanks for the info | 13:19 |
| tbarron | zaneb: kk | 13:19 |
| tbarron | on that tangent swift and manila are ok for either cause they don't have tight nova/hypervisor coupling, deliver storage over network to the container host | 13:20 |
| zaneb | tbarron: cinder does work with ironic using iscsi though, I thought? does that not count? | 13:20 |
| tbarron | zaneb: yes but cinder-csi doesn't use that approach and it's a very different code path | 13:21 |
| zaneb | ah | 13:21 |
| * zaneb will have to update the teapot docs | 13:21 | |
| tbarron | so conceptually it could be done | 13:21 |
| tbarron | and work for iscsi | 13:21 |
| tbarron | and with work probably for rbd-nbd | 13:21 |
| tbarron | rbd kernel modules lack features cinder uses with they hypervisor-path rbd | 13:22 |
| tbarron | but the code in cloud-provider-openstack isn't written to do that and it's not trivial to support both paths | 13:23 |
| zaneb | fungi: btw the thing you've heard about with using the Cinder drivers directly without going through the ReST API is called Ember CSI | 13:25 |
| fungi | zaneb: aha! yes that name sounds familiar | 13:25 |
| zaneb | but it's not multi-tenant obviously, because it's the ReST API that implements multitenancy | 13:25 |
| fungi | yep | 13:25 |
| fungi | makes sense, of course | 13:26 |
| fungi | i also keep hearing "kubernetes users don't care about multi-tenancy" so i suppose whether it's appropriate depends on your use case | 13:26 |
| zaneb | they don't care about multitenancy within a k8s deployment | 13:27 |
| fungi | (obviously at least some kubernetes providers care about tenant isolation, i suppose that's the usual user/operator split of concerns) | 13:27 |
| zaneb | I mean, they do a little, but in a different way and it's complicated | 13:27 |
| zaneb | but they expect the underlying cloud to provide multitenancy so they can run lots of clusters in different tenants in one cloud | 13:28 |
| zaneb | so in the context of defining a k8s starter-kit for OpenStack, I would expect people to use Manila CSI or maybe Cinder CSI but not Ember CSI | 13:30 |
| zaneb | Ember is useful if you have a single large on-prem bare-metal k8s cluster | 13:30 |
| *** e0ne has quit IRC | 13:37 | |
| *** e0ne has joined #openstack-tc | 13:37 | |
| fungi | got it | 13:37 |
| *** rpittau is now known as rpittau|brb | 13:39 | |
| mugsie | zaneb: block storage in K8S is used - it is how the default volume support is done in AKS. but the agent on the host just mounts the volume, and maps it into the pods namespace | 13:39 |
| mugsie | "just" varies per provider | 13:39 |
| *** ricolin has quit IRC | 13:40 | |
| openstackgerrit | Zane Bitter proposed openstack/ideas master: Teapot: Add correction about Cinder CSI support https://review.opendev.org/736220 | 13:41 |
| zaneb | tbarron: fix to teapot docs ^ | 13:41 |
| mugsie | could os-brick not run on the barememtal image to mount the volume? | 13:43 |
| mugsie | https://review.opendev.org/#/c/263744/ was added for that I think | 13:45 |
| *** kberger_ has joined #openstack-tc | 13:45 | |
| *** kberger_ has quit IRC | 13:46 | |
| zaneb | mugsie: IIUC the issue is that the CSI driver doesn't work that way | 13:46 |
| *** kberger_ has joined #openstack-tc | 13:46 | |
| tbarron | zaneb: ack | 13:47 |
| *** e0ne has quit IRC | 13:48 | |
| *** e0ne_ has joined #openstack-tc | 13:48 | |
| tbarron | zaneb: mugsie i'll add e0ne to 263744 in case i'm missing something | 13:49 |
| tbarron | love to be wrong on this one | 13:49 |
| *** KeithMnemonic has quit IRC | 13:49 | |
| *** rpittau|brb is now known as rpittau | 13:53 | |
| smcginnis | mugsie: python-brick-cinderclient-ext was created to be able to perform those operations normally done by nova. | 13:54 |
| smcginnis | So that could be used for that. I guess the only trick is to get it installed on the baremetal host. | 13:54 |
| mugsie | yeah. In a teapot usecase the image would be controlled by the operator, so requiring it to be installed isn't a huge deal | 13:55 |
| tbarron | fungi: zaneb -- on embber-csi tangent (and it is a tangent for the multi-tenancy reasons zaneb mentioned) geguileo told me he's planning to use rbd-ngd to get feature parity with hypervisor path for the baremetal case | 13:55 |
| *** lpetrut has joined #openstack-tc | 13:56 | |
| tbarron | smcginnis: mugsie yeah i think the trick is getting cloud-provider-openstack cinder-csi module to use it | 13:56 |
| zaneb | mugsie: I don't agree that in all use cases for Teapot the image would be controlled by the operator, but I agree that it's fine to say that if you want block storage you have to install both Cinder CSI and this other thing | 13:57 |
| jungleboyj | smcginnis: The python-brick-cinderclient-ext hasn't seen much love lately. | 14:00 |
| jungleboyj | I was using it for a while. Maybe it is just working? | 14:01 |
| fungi | "feature complete" ;) | 14:02 |
| smcginnis | It's hard to tell if it is being used and just working fine, or if it's not being used and if there is something missing or broken that we don't know about. | 14:06 |
| smcginnis | We have tests, and the tests pass, so my assumption is it is working, just not widely used. | 14:06 |
| openstackgerrit | Ghanshyam Mann proposed openstack/project-team-guide master: Clarify deprecation tags and record deprecation cycle https://review.opendev.org/736082 | 14:23 |
| tbarron | smcginnis: jungleboyj So I am thinking that just installing python-brick-cinderclient-ext isn't sufficient to get cinder-csi/gophercloud working for baremetal but would love to be wrong | 14:24 |
| jungleboyj | smcginnis: That is probably true. | 14:25 |
| jungleboyj | tbarron: I had installed and had it working in the past and I think you are right that there is probably more that needs to be done to make it work. It was a long time ago though. | 14:26 |
| *** dklyle has joined #openstack-tc | 14:44 | |
| *** e0ne_ has quit IRC | 15:40 | |
| *** e0ne has joined #openstack-tc | 15:46 | |
| *** e0ne has quit IRC | 16:01 | |
| *** rpittau is now known as rpittau|afk | 16:17 | |
| openstackgerrit | Ghanshyam Mann proposed openstack/governance master: Record deprecated cycle for deprecated project https://review.opendev.org/736084 | 16:38 |
| *** diablo_rojo has joined #openstack-tc | 16:58 | |
| *** lpetrut has quit IRC | 17:02 | |
| *** e0ne has joined #openstack-tc | 17:03 | |
| *** kberger_ has quit IRC | 17:20 | |
| openstackgerrit | Kendall Nelson proposed openstack/governance master: Update joining-tc.rst to be general tc-guide.rst https://review.opendev.org/732983 | 17:22 |
| mnaser | tc-members: friendly reminder on https://review.opendev.org/#/c/735361/ | 17:24 |
| mnaser | the more this waits, the slower it takes to merge :( | 17:24 |
| mnaser | as we need at least 2/3rds for that change | 17:25 |
| *** KeithMnemonic has joined #openstack-tc | 17:25 | |
| *** KeithMnemonic has quit IRC | 17:27 | |
| *** KeithMnemonic has joined #openstack-tc | 17:37 | |
| *** e0ne has quit IRC | 17:43 | |
| *** KeithMnemonic has quit IRC | 17:43 | |
| gmann | tc-members: please reply on ML if any objection on grenade testing for Extended Maintenance proposal - http://lists.openstack.org/pipermail/openstack-discuss/2020-June/015499.html | 18:19 |
| openstackgerrit | Kendall Nelson proposed openstack/governance master: Update joining-tc.rst to be general tc-guide.rst https://review.opendev.org/732983 | 18:46 |
| *** jamesmcarthur has joined #openstack-tc | 18:47 | |
| *** jamesmcarthur has quit IRC | 18:54 | |
| *** jamesmcarthur has joined #openstack-tc | 18:55 | |
| *** jamesmcarthur_ has joined #openstack-tc | 18:59 | |
| *** jamesmcarthur has quit IRC | 19:01 | |
| *** ralonsoh has quit IRC | 19:22 | |
| *** jamesmcarthur_ has quit IRC | 19:45 | |
| *** jamesmcarthur has joined #openstack-tc | 19:46 | |
| openstackgerrit | Zane Bitter proposed openstack/ideas master: Fix formatting https://review.opendev.org/736339 | 19:47 |
| openstackgerrit | Zane Bitter proposed openstack/governance master: Create starter-kit:kubernetes-in-virt tag https://review.opendev.org/736369 | 20:11 |
| *** tobberydberg_ has joined #openstack-tc | 20:12 | |
| *** tosky_ has joined #openstack-tc | 20:12 | |
| *** tosky has quit IRC | 20:13 | |
| *** tosky_ is now known as tosky | 20:13 | |
| *** irclogbot_1 has quit IRC | 20:14 | |
| openstackgerrit | Zane Bitter proposed openstack/governance master: Create starter-kit:kubernetes-in-virt tag https://review.opendev.org/736369 | 20:14 |
| *** purplerbot has quit IRC | 20:18 | |
| *** tobberydberg has quit IRC | 20:18 | |
| *** masayukig has quit IRC | 20:18 | |
| *** dtruong has quit IRC | 20:18 | |
| *** dtruong has joined #openstack-tc | 20:19 | |
| zaneb | first draft for y'all's consideration above ^ | 20:21 |
| *** irclogbot_0 has joined #openstack-tc | 20:21 | |
| *** ijolliffe has quit IRC | 20:23 | |
| *** purplerbot has joined #openstack-tc | 20:23 | |
| *** masayukig has joined #openstack-tc | 20:24 | |
| diablo_rojo | zaneb, thanks for getting that started! | 20:29 |
| zaneb | np | 20:29 |
| zaneb | feel free to suggest a better name btw. I hate all the options I thought of | 20:30 |
| *** ijolliffe has joined #openstack-tc | 20:30 | |
| diablo_rojo | Naming is sooo very hard lol | 20:31 |
| *** jamesmcarthur has quit IRC | 20:31 | |
| *** jamesmcarthur has joined #openstack-tc | 20:32 | |
| *** ildikov has quit IRC | 20:35 | |
| *** lxkong has quit IRC | 20:35 | |
| *** jamesmcarthur has quit IRC | 20:37 | |
| *** csatari has quit IRC | 20:39 | |
| *** aprice has quit IRC | 20:40 | |
| *** knikolla has quit IRC | 20:40 | |
| *** knikolla has joined #openstack-tc | 20:42 | |
| *** tonyb[m] has quit IRC | 20:43 | |
| *** aprice has joined #openstack-tc | 20:44 | |
| *** ijolliffe has quit IRC | 20:45 | |
| *** aprice has quit IRC | 20:48 | |
| *** knikolla has quit IRC | 20:52 | |
| *** diablo_rojo_phon has joined #openstack-tc | 21:00 | |
| *** e0ne has joined #openstack-tc | 21:03 | |
| *** aprice has joined #openstack-tc | 21:04 | |
| *** jamesmcarthur has joined #openstack-tc | 21:04 | |
| *** lxkong has joined #openstack-tc | 21:09 | |
| *** aprice has quit IRC | 21:10 | |
| *** noonedeadpunk has quit IRC | 21:12 | |
| *** jamesmcarthur has quit IRC | 21:13 | |
| *** lxkong has quit IRC | 21:16 | |
| *** aprice has joined #openstack-tc | 21:24 | |
| openstackgerrit | Mohammed Naser proposed openstack/governance master: Deprecatre neutron-fwaas and neutron-fwaas-dashboard master branch https://review.opendev.org/735828 | 21:28 |
| openstackgerrit | Mohammed Naser proposed openstack/governance master: Deprecatre neutron-fwaas and neutron-fwaas-dashboard master branch https://review.opendev.org/735828 | 21:29 |
| *** aprice has quit IRC | 21:30 | |
| *** ildikov has joined #openstack-tc | 21:33 | |
| *** lxkong has joined #openstack-tc | 21:34 | |
| *** aprice has joined #openstack-tc | 21:34 | |
| *** tonyb[m] has joined #openstack-tc | 21:38 | |
| *** csatari has joined #openstack-tc | 21:38 | |
| *** knikolla has joined #openstack-tc | 21:40 | |
| *** slaweq has quit IRC | 21:41 | |
| *** jamesmcarthur has joined #openstack-tc | 21:50 | |
| *** jamesmcarthur has quit IRC | 21:54 | |
| *** tosky has quit IRC | 21:56 | |
| *** tosky has joined #openstack-tc | 21:56 | |
| *** e0ne has quit IRC | 22:19 | |
| *** tkajinam has joined #openstack-tc | 22:54 | |
| *** tosky has quit IRC | 23:09 | |
| *** markvoelker has joined #openstack-tc | 23:24 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!