| opendevreview | Merged openstack/openstack-ansible-os_nova master: Ensure nova_device_spec is templated as JSON string https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/913248 | 10:24 |
|---|---|---|
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2023.2: Use ansible_facts['processor_vcpus'] instead of fact variable https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/912753 | 10:39 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2023.1: Use ansible_facts['processor_vcpus'] instead of fact variable https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/912754 | 10:41 |
| gebz_ | New to IRC, how do i connect to this channel via hexchat? | 10:49 |
| gebz97 | Anybody here? | 10:52 |
| noonedeadpunk | o/ | 11:01 |
| noonedeadpunk | gebz: hey | 11:02 |
| gebz | what's up | 11:02 |
| gebz | new here | 11:02 |
| noonedeadpunk | sure - feel free to fire your questions :) ppl are mostly around here during EU business hours though. But we can be around quite late to catch up with EST as well :) | 11:03 |
| gebz | It's ok, I'm GMT+2 | 11:03 |
| gebz | I mostly have a general idea what openstack's architecture looks like, but i'm really struggling to get wrap my head around neutron | 11:05 |
| noonedeadpunk | oh, well :D you're not the only one for sure | 11:05 |
| gebz | :D | 11:05 |
| noonedeadpunk | imo, networking is really the hardest topic for me as well | 11:05 |
| gebz | Simply put, do the neutron networks only exist as a tunneled network inside openstack? (i.e. they require BGP peering with data center core network to be visible?) | 11:06 |
| gebz | or can they exist as vlans on the DC network? | 11:07 |
| noonedeadpunk | I guess it depends on the type of network | 11:08 |
| gebz | spine-leaf | 11:08 |
| noonedeadpunk | If we're talking about tenant private networks (east-west traffic) - they exist only as vxlan or geneve tunnels (depending on the driver) | 11:09 |
| jrosser | (you can also do those as vlans if you want to) | 11:09 |
| noonedeadpunk | Or well. They can be also vlans in theory, though vlans don't scale in a good way, so usually they're not used for that | 11:09 |
| gebz | what about the openstack infrastructure hosts? | 11:10 |
| noonedeadpunk | but no, there's no requirement of having BGP visible for any networks | 11:10 |
| noonedeadpunk | Like you can do just fine with vlans for north-south just fine | 11:10 |
| gebz | but the switch ports that the physical hosts are connected to need to be trunked correct? | 11:10 |
| noonedeadpunk | well, yes, that's how vlans work? | 11:11 |
| noonedeadpunk | this is not required for vxlan/geneve though | 11:11 |
| noonedeadpunk | as it's tunneled traffic | 11:12 |
| gebz | same as kubernetes pod network/service network right? | 11:12 |
| noonedeadpunk | or well, again I guess depends | 11:12 |
| noonedeadpunk | I think pretty much yes. | 11:12 |
| noonedeadpunk | So you basically set IP/network which will be used for the tunneled traffic | 11:13 |
| noonedeadpunk | and neutron ensures to create vxlan/geneve on top of that | 11:13 |
| noonedeadpunk | but that's about east-west more or less | 11:13 |
| jrosser | those tunneled networks will go just fine over your spine/leaf between computes and infra (specifically neutron l3 agent or OVN gateway nodes) | 11:13 |
| noonedeadpunk | for north-south you'd need either just vlan, or indeed go with bgp solution | 11:13 |
| jrosser | you need to consider separately what is going to happen for external “provider” | 11:14 |
| jrosser | networks that might be l2 constructs and won’t natively get across that spine/leaf | 11:14 |
| noonedeadpunk | well, there're ways to make them.... | 11:14 |
| jrosser | indeed, but this is an important decision | 11:15 |
| noonedeadpunk | ++ | 11:15 |
| jrosser | like you can dedicate a leaf to the control plane | 11:15 |
| gebz | a leaf switch for the openstack infra? | 11:15 |
| jrosser | or you can use vxlan on the network hardware to span these things where you need them | 11:15 |
| * noonedeadpunk had a partial fiasco now with ovn-bgp-agent | 11:15 | |
| gebz | like a top-of-rack? | 11:15 |
| jrosser | gebz: so for example, in my largest deployment I have a pair of switches dedicated to the control plane | 11:16 |
| jrosser | but that’s just because the design I have needed that number of ports | 11:16 |
| opendevreview | Merged openstack/openstack-ansible master: Bump ansible version to 2.15.9 https://review.opendev.org/c/openstack/openstack-ansible/+/905619 | 11:16 |
| gebz | but those 2 switches is where for example the management network would live, correct? | 11:17 |
| gebz | also the storage network | 11:18 |
| jrosser | the management network has to go to all the nodes | 11:18 |
| gebz | Ok, the network team's gonna be more involved in this than I originally thought :'D | 11:18 |
| jrosser | so your choice would be to have it be l3 on your leaf/spine | 11:18 |
| jrosser | or you do it l2 with evpn or something | 11:19 |
| jrosser | or of course l3 inside evpn is also possible | 11:19 |
| jrosser | I think that what I’m saying is the logical architecture is important to tie down rather than just looking at the pile of hardware and deciding how to plug it up | 11:19 |
| gebz | Right now I'm sort of in a PoC phase.. trying to pitch openstack to our CTO, doing the deployment inside of proxmox | 11:20 |
| gebz | We're 97% VMWare atm | 11:20 |
| jrosser | obviously everything I have said is entirely dependant on what scale you might want to achieve | 11:20 |
| jrosser | smaller things can be much much simpler | 11:21 |
| gebz | 3 physical hosts with proxmox | 11:21 |
| gebz | 120 vCPUs, 0.5 TiB ram, ~200 TiB storage (40TiB NVME/160TiB HDD) | 11:22 |
| gebz | total | 11:22 |
| gebz | Those are the boxes i'm working with rn | 11:22 |
| jrosser | well “I want to build a small openstack and already have a spine/leaf network” is not the same as “I want to build an openstack so large that a spine/leaf network is the only way” | 11:23 |
| gebz | all are on the same vlan, connected to the same top of rack | 11:23 |
| gebz | no, i meant there was already a spine/leaf in our data center | 11:23 |
| gebz | Must have misunderstood :D | 11:23 |
| gebz | Is there any real difference if I have the infra hosts as VMs instead of on bare metal? | 11:29 |
| noonedeadpunk | no, not really, except performance conciderations | 11:31 |
| noonedeadpunk | eventually, you can also split each component to it's own VM easily | 11:31 |
| gebz | not too big an issue since it would be a PoC | 11:31 |
| noonedeadpunk | you can actually just spawn everything in a 1 single VM including ceph, through AIO build | 11:32 |
| gebz | That sounds interesting actually.. | 11:33 |
| gebz | In that case, does that chunky boi needs to have 6x VNICs? | 11:34 |
| gebz | with 6 IP addresses? | 11:34 |
| noonedeadpunk | https://docs.openstack.org/openstack-ansible/latest/user/aio/quickstart.html | 11:34 |
| noonedeadpunk | No, just 1 VM with 16gb of ram, 4-8 vcpus and 100gb of disk | 11:34 |
| noonedeadpunk | 1 nic is fine | 11:34 |
| noonedeadpunk | rest will be "dummied" | 11:34 |
| noonedeadpunk | actually, there's even eaasier way of doing All-In-One then in the doc... | 11:35 |
| gebz | Tell me | 11:35 |
| noonedeadpunk | basically - git clone https://opendev.org/openstack/openstack-ansible; cd openstack-ansible; ./scripts/gate-check-commit.sh aio_lxc_ceph` | 11:35 |
| gebz | lol that simple? | 11:36 |
| noonedeadpunk | you can add any service to the list of `aio_lxc_ceph`, ie `aio_lxc_ceph_magnum_trove_ceilometer` | 11:36 |
| noonedeadpunk | you can also replace `lxc` with `metal` if you don't wanna have LXC containers | 11:36 |
| gebz | I would be lying if I said I understand LXC containers either, mostly worked with docker and k8s | 11:37 |
| noonedeadpunk | they're just system containers | 11:37 |
| noonedeadpunk | not application ones like docker | 11:37 |
| noonedeadpunk | so consider them as a very lightweight VM with a shared kernel | 11:38 |
| noonedeadpunk | Like OpenVZ :D | 11:38 |
| gebz | F**k it imma do it right now | 11:38 |
| noonedeadpunk | we don't do docker or k8s here, but there're deployment projects who do | 11:39 |
| gebz | I see | 11:40 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron stable/2023.1: Use ansible_facts['processor_vcpus'] instead of fact variable https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/912754 | 11:40 |
| gebz | Ok, created a rhel vm with 16 vcpu, 32G ram, and 100G nvme, let's see what this bad boy can do :D | 11:49 |
| andrewbonney | noonedeadpunk: have you seen any issues with reply queues when taking down and rebuilding rmq nodes? | 11:56 |
| andrewbonney | These only live on one node in a set due to https://github.com/openstack/openstack-ansible-rabbitmq_server/commit/52ad552129afc715dc978c61edf881090fcf48c0 and we've seen issues with them not getting re-created during maintenance | 11:56 |
| noonedeadpunk | andrewbonney: frankly speaking, I'm not absolutely sure here. Like we totally had an issue with heat-engine - each time it was loosing connection to rabbitmq it needed a restart more or less | 12:00 |
| noonedeadpunk | gebz: I'm not 100% sure about health of ceph deploy on rhel today... bit worth trying/fixing. we jsut don't have that job in our CI | 12:01 |
| gebz | debian/ubuntu/suse? | 12:02 |
| noonedeadpunk | ubuntu works 100% | 12:03 |
| gebz | rhel is the simplest to manage because im on a restricted network | 12:03 |
| noonedeadpunk | debian should work, so that should rhel as well. | 12:03 |
| noonedeadpunk | eventually, we should catch/fix that if it doesn't, but we just don't run any EL in productions, so I don't really know if today it works | 12:04 |
| noonedeadpunk | worth trying :) | 12:04 |
| gebz | i forked stable/2023.2 | 12:04 |
| gebz | cloned* | 12:04 |
| noonedeadpunk | btw, we have whole section about limited connectivity... | 12:04 |
| gebz | noticed :D | 12:05 |
| gebz | spent 3 hours yesterday spinning in circles till i found it | 12:05 |
| gebz | Documentation team ought to have that at the beginning | 12:05 |
| noonedeadpunk | contributions are always welcome :D | 12:05 |
| gebz | as "special considerations" section in the introduction or something | 12:06 |
| gebz | is there a repo for the documentation? | 12:06 |
| noonedeadpunk | well, we have it a bit spreaded. So we have docs for the project, and it's stored here: https://opendev.org/openstack/openstack-ansible/src/branch/master/doc/source | 12:07 |
| noonedeadpunk | Then there's a deploy guide, which is a bit different instance: https://opendev.org/openstack/openstack-ansible/src/branch/master/deploy-guide/source | 12:08 |
| noonedeadpunk | and then we also have document pages for each role separately, descriping possible options/usecases | 12:08 |
| noonedeadpunk | ie - for neutron: https://docs.openstack.org/openstack-ansible-os_neutron/latest/ | 12:08 |
| gebz | ok, my finger is on the trigger, you said ./scripts/gate-check-commit.sh aio_lxc_ and what else can i stuff in there? :D | 12:09 |
| noonedeadpunk | which is basically here: https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/doc/source | 12:09 |
| noonedeadpunk | gebz: pretty much wahtever project you want to see installed | 12:09 |
| gebz | how may are there? | 12:09 |
| gebz | :DD | 12:09 |
| noonedeadpunk | aio - stands for just basics - keystone, horizon, placement, nova, cinder, glance, neutron | 12:09 |
| gebz | does aio include swift? | 12:11 |
| noonedeadpunk | so basically anything from this list, cutting `os_` preffix: https://docs.openstack.org/project-deploy-guide/openstack-ansible/latest/configure.html#openstack-service-roles | 12:11 |
| noonedeadpunk | no, it does not | 12:11 |
| noonedeadpunk | but. | 12:11 |
| noonedeadpunk | ceph does | 12:11 |
| noonedeadpunk | or well | 12:11 |
| gebz | oh yea | 12:12 |
| noonedeadpunk | I guess there's a choice you'd need to make, if you wanna to have swift-swift or RGW offering Swift-compatible API | 12:12 |
| gebz | CTO was more interested in ceph, so lets do ceph | 12:12 |
| gebz | I pressed the big red button :D | 12:16 |
| gebz | It failed in the middle because it tried to ssh to itself using the new IPs but they weren't in the known_hosts | 12:33 |
| gebz | fixed that, reran, now it failed at bootstrap-host create LVM LV | 12:34 |
| gebz | I guess it doesn't like the fact that a ceph cluster was already created | 12:34 |
| gebz | is there a way to clean it up and rerun? | 12:35 |
| gebz | I see what you mean why it might fail on rhel, no repo called crb.. | 12:46 |
| noonedeadpunk | huh | 13:04 |
| noonedeadpunk | sorry, was in a meeting | 13:04 |
| noonedeadpunk | so, it would be really interesting to see an output of the first failure. | 13:04 |
| noonedeadpunk | as it's super weird, given that in ansible settings it should ignore known_hosts for that kind of build | 13:05 |
| noonedeadpunk | https://opendev.org/openstack/openstack-ansible/src/branch/master/scripts/openstack-ansible.rc#L48 | 13:06 |
| noonedeadpunk | gebz: regarding crb repo - can you please paste output of the task that failed? | 13:06 |
| noonedeadpunk | like - thorugh https://paste.openstack.org/ or smth? | 13:07 |
| gebz | I could rerun it again later | 13:15 |
| noonedeadpunk | I wonder if that's smth available for centos/rocky but not for RHEL directly | 13:15 |
| noonedeadpunk | as we don't have RHEL anywhere handy to test things against it | 13:15 |
| gebz | Thing is, Im using satellite to manage local repos | 13:16 |
| gebz | might be a reason, hold on i will rerun it | 13:16 |
| gebz | machine won't boot :'D | 13:19 |
| gebz | might be some systemd service that's stuck | 13:19 |
| gebz | ah.. networkmanager got disabled | 13:20 |
| gebz | rhel9 uses networkmanager by default | 13:20 |
| jrosser | gebz: this is known to work on rocky9 so it should be very similar | 13:21 |
| jrosser | and the all-in-one uses systemd-networkd to manage its networking | 13:21 |
| gebz | Is rocky 9 officially supported by the project? | 13:23 |
| jrosser | yes, because it is possible to test that in our CI | 13:23 |
| jrosser | which is not the case for RHEL | 13:23 |
| gebz | Perfect | 13:23 |
| gebz | gonna tear it down and rebuild :D | 13:23 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/xena: Switch SHAs to EOM https://review.opendev.org/c/openstack/openstack-ansible/+/913413 | 15:28 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/wallaby: Switch SHAs to EOM https://review.opendev.org/c/openstack/openstack-ansible/+/913414 | 15:33 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/victoria: Switch SHAs to EOM https://review.opendev.org/c/openstack/openstack-ansible/+/913418 | 15:36 |
| noonedeadpunk | jrosser: have you seen that there's presumably an OAUTH support now in Skyline? | 15:37 |
| noonedeadpunk | https://review.opendev.org/c/openstack/kolla-ansible/+/905860 | 15:38 |
| jrosser | noonedeadpunk: i saw there was something but there was basically no docs | 15:38 |
| noonedeadpunk | I've seen they're doing some l18n attempts as well | 15:39 |
| jrosser | you see the thing is that openid != openid-connect | 15:39 |
| noonedeadpunk | (which is nice for non-EN I guess :D) | 15:39 |
| noonedeadpunk | well.... that is true.... | 15:39 |
| gebz | retried building an AIO stack on rocky 9 | 15:40 |
| jrosser | and unfortunately sometimes these get used interchangeably | 15:40 |
| jrosser | but everything i see in skyline only refers to openid | 15:40 |
| gebz | This time it failed at LXC installation, proxy flagged the repo as suspicious :'D | 15:41 |
| jrosser | and so i never see any place to configure client-id / client-secret like you'd have with google auth or whatever | 15:41 |
| noonedeadpunk | gebz: ah, well... true... I think we're using a NeilHanlon's one at the moment specifically for templates? | 15:48 |
| gebz | yup | 15:48 |
| gebz | I looked at it, only 40 downloads :'D seems risky even to me | 15:49 |
| gebz | Im redoing it on ubuntu, passed the LXC step | 15:53 |
| gebz | been on the "build the base image using CLI tool" for ~20 mins now | 15:53 |
| jrosser | ^ this should be quick, just a minute or two | 15:54 |
| noonedeadpunk | gebz: you could try just metal, without LXC | 15:54 |
| gebz | does it download anything external at this step? | 15:54 |
| noonedeadpunk | it does, yes | 15:54 |
| gebz | ah.. I see... | 15:54 |
| gebz | must be the damn proxy again | 15:54 |
| noonedeadpunk | it's building the image using debootrstrap | 15:54 |
| gebz | ^ I'll give that a try after I've had something to eat, been at this for hours :'D | 15:56 |
| jrosser | if you have a complicated environment with proxies and limited connectivity, it's going to take some time and effort to get things working | 15:56 |
| jrosser | this is not a "shrink wrap" type installation | 15:57 |
| noonedeadpunk | yeah, so that's a repo build by NeilHanlon for us, so that we could have lxc-templates-extra package | 15:57 |
| gebz | @jrosser tell me about it :'D | 15:57 |
| noonedeadpunk | I can recall smth was changed in epel regarding LXC, so maybe it's not needed anymore... but not sure | 15:57 |
| noonedeadpunk | huh, just found https://github.com/ganto/copr-lxc4/tree/master | 15:58 |
| noonedeadpunk | so we can be able now to replace that one.... | 15:58 |
| jrosser | gebz: the idea is that all the hooks are there, so if you have a local repo mirror, you can point to it | 15:58 |
| jrosser | or if you locally cache pypi packages, you can point to that too | 15:59 |
| noonedeadpunk | and seems they've also builded incus | 15:59 |
| jrosser | but we don't provide tooling to create any of that, as every place has pretty much their own requirements | 15:59 |
| noonedeadpunk | but that's only for fedora /o\ | 16:00 |
| jrosser | also http proxies are pretty well supported during the deployment, but you need to understand and take care with what you are doing | 16:00 |
| gebz | Only got satellite on-prem, and that's for rhel.. no repo mirrors for other distros | 16:03 |
| jrosser | well, if you have an http proxy it should work | 16:04 |
| noonedeadpunk | I wonder if we still need lxc-templates-extras..... | 16:05 |
| ThiagoCMC | jrosser, nobody is replying to the ceph-ansible GitHub issue (https://github.com/ceph/ceph-ansible/issues/7496)... I'm thinking about sending an e-mail to the guy from IBM that's changing the `stable-8.0` branch. Do you think worth contacting him? If yes, want me to including any message from you? | 16:14 |
| jrosser | ThiagoCMC: well they did put back the config overrides bit in a later patch | 16:17 |
| jrosser | so you might want to edit that a bit | 16:17 |
| jrosser | and it's also likley that the data that was taken out specifically for openstack can be moved instead into osa group_vars | 16:19 |
| ThiagoCMC | Oh, nice! Thanks for the info! The OpenStack-related bits you're talking about are these: https://github.com/ceph/ceph-ansible/commit/9c467e41b39af63dec9a0b237b550edb9c57da85 - Right? | 16:21 |
| jrosser | ThiagoCMC: i have not checked in detail | 16:25 |
| ThiagoCMC | Ok | 16:26 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Use container setup role from plugins repo https://review.opendev.org/c/openstack/openstack-ansible/+/905004 | 16:38 |
| noonedeadpunk | jrosser: huh, seems you did 2 patches?:) https://review.opendev.org/c/openstack/openstack-ansible/+/905004 vs https://review.opendev.org/c/openstack/openstack-ansible/+/908984 | 16:40 |
| noonedeadpunk | which you wanna leave? :LD | 16:40 |
| noonedeadpunk | I'd vote for one not in merge conflict :D | 16:41 |
| jrosser | oh hah | 17:01 |
| opendevreview | Merged openstack/openstack-ansible-os_neutron stable/2023.1: Use ansible_facts['processor_vcpus'] instead of fact variable https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/912754 | 17:31 |
| NeilHanlon | noonedeadpunk, gebz: yeah .. the lxc-templates stuff is (as I understand) deprecated and shouldn't be used. I don't even know that we need it any longer as I think we changed how we are creating the containers | 17:43 |
| noonedeadpunk | yeah... | 17:43 |
| noonedeadpunk | worth checking that... | 17:43 |
| gebz | :D | 17:43 |
| gebz | I'M HELPING | 17:43 |
| NeilHanlon | and apologies for being suspicious :D | 17:43 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_horizon master: Add support for deploying Horizon with Skyline https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/913439 | 17:54 |
| noonedeadpunk | crap | 18:35 |
| noonedeadpunk | seems skyline does not support running on sub-urls | 18:35 |
| noonedeadpunk | ie /skyline | 18:35 |
| NeilHanlon | but... why | 18:40 |
| NeilHanlon | i guess we can rewrite the urls in haproxy but..... why | 18:40 |
| noonedeadpunk | NeilHanlon: I'm sooooo long ago was dealing with web server config | 18:49 |
| noonedeadpunk | so help is appreciated | 18:49 |
| noonedeadpunk | so the thing is, they do have this in their static file: https://opendev.org/openstack/skyline-console/src/branch/master/skyline_console/static/index.html#L11 | 18:49 |
| noonedeadpunk | so we need to kinda rewrite all uris to contain /skyline? | 18:49 |
| NeilHanlon | ew.. | 18:49 |
| noonedeadpunk | hm. maybe you're right, rewrite is exactly what I need... | 18:50 |
| NeilHanlon | possible, but also.. those files shouldn't be in git.. | 18:51 |
| noonedeadpunk | lol | 18:54 |
| noonedeadpunk | yeah. static files are really static :D | 18:54 |
| noonedeadpunk | fwiw, this repo contains _only_ such static files. Which are installable through python :D | 18:55 |
| NeilHanlon | yep.. so i think they actually get overwritten | 18:56 |
| NeilHanlon | https://opendev.org/openstack/skyline-console/src/branch/master/config/webpack.prod.js | 18:56 |
| noonedeadpunk | hm | 18:59 |
| NeilHanlon | i *think* that the index.html in there needs to be modified to support webPack putting in the right info.. https://github.com/jantimon/html-webpack-plugin#options | 19:01 |
| NeilHanlon | but I'm also not a web guy anymore lol | 19:01 |
| NeilHanlon | are we rebuilding ? or just installing from pip | 19:01 |
| NeilHanlon | i was assuming we were running the build in repo_container but I guess that is a bad assumption | 19:02 |
| noonedeadpunk | I think we just installing | 19:02 |
| noonedeadpunk | well | 19:02 |
| noonedeadpunk | building wheels is not helpful to adjust it I assume | 19:02 |
| noonedeadpunk | but yes, we build first kinda | 19:03 |
| noonedeadpunk | though we don't trigger npm or anything like that I assume | 19:03 |
| NeilHanlon | i think not, as they bundle the static content before building the wheel https://opendev.org/openstack/skyline-console/src/branch/master/Makefile#L77 | 19:03 |
| NeilHanlon | otoh, if the url in their template wasn't using `/` at the start, this would just work, i think | 19:04 |
| noonedeadpunk | I guess so as well, yes | 19:11 |
| NeilHanlon | sed the file? 😂 | 19:13 |
| noonedeadpunk | rewrite actually results in cycle redirect.... | 19:13 |
| noonedeadpunk | yeah, so dunno | 19:13 |
| NeilHanlon | i'll try to poke at your change this weekend | 19:26 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [Feature] Add skyline deployment capability https://review.opendev.org/c/openstack/openstack-ansible/+/859446 | 19:30 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: [Feature] Add skyline deployment capability https://review.opendev.org/c/openstack/openstack-ansible/+/859446 | 19:31 |
| noonedeadpunk | I think we can consider this OK as long as we're OK with Skyline being default, when it's installed together with Horizon | 19:32 |
| noonedeadpunk | While Horizon will be under `/horizon` | 19:32 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_horizon master: Reflect horizon_webroot setting in Apache vhost configuration https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/913442 | 19:36 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_horizon master: Add support for deploying Horizon with Skyline https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/913439 | 19:38 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_horizon master: Add support for deploying Horizon with Skyline https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/913439 | 19:38 |
| jrosser | noonedeadpunk: NeilHanlon if you look at the very first version of my os_skyline role (perhaps the one in my github early commits) i did the whole npm build from scratch | 23:32 |
| jrosser | it was very slow and needed a gigantic amount of RAM, many GB, so would have never worked in a CI job | 23:33 |
| NeilHanlon | yeah.. that sounds like webpack, honestly | 23:58 |
| * NeilHanlon grumbles about javascript | 23:59 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!