| tonyb | ironic folks, it looks like https://review.opendev.org/c/openstack/requirements/+/949342 will merge "real soon now". That means that *next weekend* when we automatically update constraints we'll possibly? pull in bcrypt===5.0.0 Have you done any testing on that? | 05:41 |
|---|---|---|
| *** JoyAkinyi is now known as Guest29989 | 07:07 | |
| rpittau | good morning ironic! o/ | 08:22 |
| Ayo[m] | Hello @Pointbr8ker | 08:23 |
| Ayo[m] | To submit your task on outreachy go the project info on the project and you should see a submit a apply to this project button | 08:23 |
| Ayo[m] | Tho the project is closed so I don’t know if that’s still possible | 08:23 |
| Ayo[m] | Good morning ironic :) | 08:24 |
| Ayo[m] | Good morning rpittau :) | 08:24 |
| rpittau | hey Ayo[m] :) | 08:24 |
| Ayo[m] | Still eagerly waiting for update on the project :) | 08:25 |
| Ayo[m] | s/submit a/“/, s//”/ | 08:26 |
| rpittau | Ayo[m]: we've started reviewing the submissions, we'll have news before the end of the week, I think :D | 09:31 |
| dtantsur | Hey folks, I'm back one week later (thanks summit covid). | 09:49 |
| frickler | tonyb: I think the best way to check would be to do a dedicated change for that in u-c and then test that in ironic with a depends-on change? or are you thinking about people testing the bump locally? | 10:02 |
| tonyb | frickler: I can do that but I figured I'd ask. as I was about to log off for the day and figured that maybe others had already thought about it | 10:06 |
| frickler | tonyb: yes, I'm not saying that you should do that yourself, more like a hint for ironic contributors. but you ping is helpful, I think the default action for everyone is just to rely on what we get from upper-constraints | 10:15 |
| zigo | Does anyone think my open bug here: https://bugs.launchpad.net/ironic/+bug/2129690 needs discussion ? Or should I just dive into implementation ? If so, I may need a bit of help. Could anyone guide me into what I should look into modifying ? I'm not familliar at all with the Ironic code base, but with a few pointers, it should be easy. Where's the source code that recieves the rabbitmq message and writes down the Grub config ? Where's the part | 10:27 |
| zigo | that is sending it ? | 10:27 |
| opendevreview | Aykut Bulgu proposed openstack/sushy stable/2025.2: Check required credentials in a detailed way https://review.opendev.org/c/openstack/sushy/+/964871 | 10:44 |
| opendevreview | Aykut Bulgu proposed openstack/sushy stable/2025.1: Check required credentials in a detailed way https://review.opendev.org/c/openstack/sushy/+/964872 | 11:14 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: [WIP] Remove tinyipa support https://review.opendev.org/c/openstack/bifrost/+/964404 | 11:22 |
| iurygregory | good morning ironic | 11:26 |
| iurygregory | PTG starts today 13 - 17 UTC Mitaka Room https://ptg.opendev.org/ptg.html Etherpad https://etherpad.opendev.org/p/ironic-ptg-2026.1 | 11:27 |
| opendevreview | Merged openstack/bifrost master: Clarify correct usage of 'install --testenv' in bifrost-cli docs https://review.opendev.org/c/openstack/bifrost/+/964461 | 11:37 |
| TheJulia | good morning | 11:58 |
| dtantsur | morning TheJulia | 11:59 |
| alegacy | Could anyone spare a moment to see what's up with this gate job (ironic-tempest-ipa-wholedisk-direct-multinode) on my review (https://review.opendev.org/c/openstack/networking-generic-switch/+/962037). It is also failing over here (https://review.opendev.org/c/openstack/networking-generic-switch/+/957207 | 12:08 |
| TheJulia | zigo: So we've had this idea come up before, and so I put a some context and thoughts into the bug. I think we even had a high level spec for the higher level rpc/interaction challenge back in 2022, but reality was we already had a pattern to call other conductors all along hiding in the code. I've pointed you to that and kind of given you some basic ideas/framing. Happy to discuss more. | 12:12 |
| TheJulia | alegacy: ugh, at a high level I know that job log will need to be dug into. Most likely won' be abel to do that until later today | 12:15 |
| opendevreview | Massimiliano Favaro-Bedford proposed openstack/ironic master: Add generic redfish compatible vendor name https://review.opendev.org/c/openstack/ironic/+/964877 | 12:16 |
| TheJulia | zigo: a key aspect is the overall "need to do pxe work" is not buffered between the core of the conductor and the write-out of the files/config via the message bus, so you need to insert a fallback call in the flow to do the needful you want to achieve. | 12:18 |
| TheJulia | zigo: oh, and there are some pxe config generation/template update calls, if you haven't found them yet just let me know | 12:19 |
| zigo | TheJulia: Thanks, that's very helpful. I'll look into all of your comments and get back with something hopefully this week (not sure if it's going to be this week: I'm on Infomaniak infra-watch duty this week, plus the vPTG...). | 12:26 |
| TheJulia | Understand completely, I'm still fighting something, although now after seeing another comment from earlier, I think I should break out a covid test | 12:29 |
| dtantsur | May be worth it, unfortunately :( | 12:52 |
| * dtantsur has just tested negatively for the first time but is still trying to cough his lungs out | 12:52 | |
| TheJulia | given how much you were masking, I'm super surprised | 12:54 |
| TheJulia | We wore masks on our flights | 12:54 |
| *** JoyAkinyi is now known as Guest30008 | 12:56 | |
| TheJulia | Summer is also practically a human covid detector now after her initial infection years ago | 12:57 |
| iurygregory | PTG starting in ~3min | 12:57 |
| iurygregory | just a headsup I will probably only be available in the first 3hrs of the PTG today, since I have two docs appt | 12:59 |
| TheJulia | ok | 13:00 |
| TheJulia | Lets get this party started! | 13:00 |
| JayF | I'm running a couple minutes behind | 13:01 |
| opendevreview | Massimiliano Favaro-Bedford proposed openstack/ironic master: Add generic redfish compatible vendor name https://review.opendev.org/c/openstack/ironic/+/964877 | 13:25 |
| opendevreview | Massimiliano Favaro-Bedford proposed openstack/ironic master: Add generic redfish compatible vendor name https://review.opendev.org/c/openstack/ironic/+/964877 | 13:49 |
| JayF | in recess until 1410 utc | 14:04 |
| opendevreview | Verification of a change to openstack/ironic stable/2024.2 failed: More RAM for Anaconda job https://review.opendev.org/c/openstack/ironic/+/964810 | 15:01 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: Remove old entries from how to release https://review.opendev.org/c/openstack/ironic/+/964891 | 15:37 |
| opendevreview | Verification of a change to openstack/ironic stable/2024.2 failed: More RAM for Anaconda job https://review.opendev.org/c/openstack/ironic/+/964810 | 15:51 |
| opendevreview | Clif Houck proposed openstack/ironic master: WIP: Generate network plan based on trait based networking config https://review.opendev.org/c/openstack/ironic/+/964895 | 15:51 |
| *** JoyAkinyi is now known as Guest30024 | 16:38 | |
| mnasiadka | Hello Ironic - is there a way to re-inspect the node maybe using the node servicing path? I’d like to very much re-inspect a node, but not really undeploy it. | 17:17 |
| TheJulia | Not really, tbh. There is or at least was a concept in inspector of saving introspection data from a node in active state, but inspection is such a different flow its not really wired in that way. | 17:27 |
| mnasiadka | Ok, thanks | 17:36 |
| TheJulia | Well "[kolla][kolla-ansible][kayobe] Bifrost usage poll" is an interesting mailing list topic | 17:36 |
| JayF | Here's a swing at rewriting the mission https://www.irccloud.com/pastebin/RtjsDyxl/projects-mission.diff.txt | 17:38 |
| mnasiadka | Yeah well, Kolla usage of Bifrost in a multiprocess single container is what we want to skip building, so either we rewire it to use Kolla-Ansible roles to deploy Ironic standalone, or we move that functionality to Kayobe (if no Kolla-Ansible users need that). | 17:39 |
| TheJulia | JayF: maybe s/To produce// ? | 17:42 |
| TheJulia | Difference of apsirational versus reality, I guess | 17:42 |
| JayF | TheJulia: the "To provide..." is the typical wording for those missions across the board | 17:49 |
| TheJulia | okay :( | 17:49 |
| JayF | e.g. nova is | 17:50 |
| JayF | To implement services and associated libraries to provide massively scalable, on demand, self service access to compute resources, including bare metal, virtual machines, and containers. | 17:50 |
| cardoe | mnasiadka: I'm interested in re-inspection via the service path as well. I was going to look at potentially making the redfish inspection not interruptive to the deployment. | 19:03 |
| opendevreview | cid proposed openstack/ironic master: Remove iLO/iLO5 driver https://review.opendev.org/c/openstack/ironic/+/965009 | 20:06 |
| opendevreview | cid proposed openstack/ironic master: Add upgrade check for iLO/iLO5 driver deprecation https://review.opendev.org/c/openstack/ironic/+/965014 | 20:27 |
| opendevreview | cid proposed openstack/ironic master: Remove iLO/iLO5 driver https://review.opendev.org/c/openstack/ironic/+/965009 | 20:30 |
| cardoe | TheJulia: around? | 20:53 |
| TheJulia | semi, whats up? | 20:55 |
| TheJulia | cardoe: ^ | 20:56 |
| cardoe | So trying to get the network stuff as least discussable. | 20:56 |
| cardoe | I wanted to get your take on https://review.opendev.org/c/openstack/ironic/+/964570 | 20:57 |
| cardoe | So to give you a concrete case... Since I think dtantsur pointed out we cannot replicate the data locally cached on a conductor pair for HA right now without shared disk I'm trying to make a little demo of all this. | 20:58 |
| TheJulia | Huh?! | 20:59 |
| cardoe | So 1 conductor for a couple of cabinets of hardware. Those boxes I'll say are on fabricDOUG-TEST which is a leaf-spine setup with redundant switches in each cabinet connecting to each port on the boxes. | 20:59 |
| cardoe | So a portgroup can be created for each server or not. | 20:59 |
| cardoe | We'll call the first cabinet cabA. | 21:00 |
| TheJulia | okay | 21:01 |
| cardoe | So I'm setting the first port on the first server in cabA with local_link_info switch_info=cabA-1, port_id=Ethernet1/1. And the second port as switch_info=cabA-2, port_id=Ethernet1/1 | 21:01 |
| cardoe | Making sense or no? | 21:01 |
| TheJulia | makes sense, *although* depending on the switch there may need to be a pre-configured port channel | 21:01 |
| TheJulia | but go on | 21:01 |
| cardoe | So then the patch I linked above is added into Ironic. And on both those ports I set the physical_network to "fabricDOUG-TEST:cabA". | 21:02 |
| cardoe | I patch Neutron to allow a physical_network name for a vxlan network. | 21:02 |
| cardoe | I then create a network segment range for type=vxlan physical_network=fabricDOUG-TEST of 10000-20000 | 21:03 |
| cardoe | I then create a network segment range for type=vlan physical_network=cabA of 1000-2000 | 21:03 |
| cardoe | I create a tenant network of type=vxlan and who cares what VNI it gets but let's just say 11000. | 21:04 |
| TheJulia | okay | 21:04 |
| TheJulia | I'm sort of following, but I know the general model and flow | 21:04 |
| cardoe | Now the little example ML2 mechanism for the Neutron folks comes into play (and I'll link it in a sec) | 21:04 |
| cardoe | It sees the physical_network of "fabricDOUG-TEST:cabA" and splits on the : and create a dynamic vlan with physical_network "cabA" | 21:05 |
| cardoe | What I'm lost on is you've wanted to also bind directly to vxlan instead of having it bind to vlan. | 21:05 |
| TheJulia | no, I've always mentally modeled on vtep -> vlan -> port | 21:06 |
| cardoe | How could I convey that hinting somehow? Would that be an admin only request? | 21:06 |
| cardoe | ah okay. So yes that's what I'm doing there is vtep -> vlan -> port. | 21:06 |
| TheJulia | The huge fundimental issue which is sort of issue two steps removed, is all the cross-node tunneling for vxlan which has to also be programmed/managed | 21:07 |
| cardoe | Absolutely. | 21:07 |
| cardoe | So the alternative bgp-agent that's using LRR handles some of that. Because it can advertise that VNI into the fabric of all the switches via BGP | 21:08 |
| cardoe | And then NGS only needs to perform the direct port binding plus the little snippet of vtep -> vlan binding. | 21:09 |
| TheJulia | LRR in this context is? | 21:10 |
| cardoe | Sorry. FRR | 21:12 |
| cardoe | https://github.com/toreanderson/evpn_agent/ | 21:12 |
| TheJulia | okay, announcing the route via BGP doesn't address the vni, only the IP level routing to enable the tunnel to be configurable and pass through | 21:15 |
| TheJulia | the VNI still needs to get configured with definitions for peer hosts, aiui | 21:15 |
| * TheJulia reads | 21:17 | |
| TheJulia | hmm, that might just work | 21:20 |
| TheJulia | per vendor testing required, for sure | 21:20 |
| cardoe | sorry. wife called to facetime with the family cause she's out of town | 22:04 |
| cardoe | yes the VNI still needs to get configured with definitions for the peer hosts, absolutely. | 22:05 |
| cardoe | But thinking we can put some of this into a NGS call | 22:05 |
| opendevreview | Julia Kreger proposed openstack/ironic master: docs: provide a little more clarity/guidance w/r/t redfish https://review.opendev.org/c/openstack/ironic/+/965030 | 22:10 |
| TheJulia | Oh yeah, that was sort of the mental model I was thinking | 22:11 |
| TheJulia | honestly, I think what makes some of the issues on the toc indexes and the resulting overload, is the fact we've got the depth enabled | 22:13 |
| TheJulia | so then it sort of a wall of words | 22:14 |
| opendevreview | cid proposed openstack/ironic master: Add upgrade check for iLO/iLO5 driver deprecation https://review.opendev.org/c/openstack/ironic/+/965014 | 22:31 |
| opendevreview | Julia Kreger proposed openstack/ironic master: docs: add a "i want a pony" bit of text https://review.opendev.org/c/openstack/ironic/+/965035 | 22:31 |
| cardoe | Am I allowed to put arbitrary fields in local_link_information? Just thinking rather than doing some kind of split on : maybe I add another field that is your vtep network. | 23:13 |
| cardoe | fabricDOUG-TEST in the above example. | 23:14 |
| cardoe | So physical_network would be the switch pair for the port group. | 23:14 |
| opendevreview | Harald Jensås proposed openstack/sushy-tools master: Add get_processors() to AbstractSystemsDriver https://review.opendev.org/c/openstack/sushy-tools/+/965036 | 23:29 |
| TheJulia | alegacy: figured it out, patch incoming | 23:43 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: Extend default timeouts https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/965037 | 23:47 |
| TheJulia | err, that one won't do it and its not long enough | 23:50 |
| TheJulia | doh | 23:50 |
| opendevreview | Harald Jensås proposed openstack/sushy-tools master: Expose System sub-resources based on driver support https://review.opendev.org/c/openstack/sushy-tools/+/965038 | 23:55 |
| opendevreview | Julia Kreger proposed openstack/ironic master: CI: make multinode runtime check sane https://review.opendev.org/c/openstack/ironic/+/965039 | 23:56 |
| opendevreview | Julia Kreger proposed openstack/ironic master: CI: make multinode runtime check sane https://review.opendev.org/c/openstack/ironic/+/965039 | 23:56 |
| TheJulia | okay ^^^ that should fix the failing multinode jobs. | 23:56 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: Extend default timeouts https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/965037 | 23:59 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!