rpittau | good morning ironic! o/ | 07:03 |
---|---|---|
*** srelf_ is now known as Continuity | 08:02 | |
Continuity | Morning Ironic o/ | 08:02 |
JayF | Do we have any plans currently to hook up the existing ilo driver (not redfish) to firmware interface? | 09:23 |
JayF | I wonder if it's as simple as plumbing the redfish support into the ilo driver for some hardware | 09:23 |
JayF | FYI gate is currently broken | 09:35 |
JayF | I put https://review.opendev.org/c/openstack/ironic/+/920365 up on Thursday to fix one issue, it was not reviewed | 09:35 |
JayF | and it appears we have some kinda pip issue preventing pep8 from running in the gate, I'm looking into that now | 09:36 |
JayF | I'm going to stack a patch on that to fix pep8 | 09:38 |
JayF | actually making it one patch or else it won't merge lol | 09:38 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Fix CI issues in codespell/pep8 https://review.opendev.org/c/openstack/ironic/+/920365 | 09:39 |
JayF | iurygregory: dtantsur: TheJulia: please someone land ^ when you are working o/ | 09:39 |
* JayF waves from London | 09:39 | |
rpittau | JayF: I'm having a look now | 09:40 |
JayF | oh hey, that's right, you're in my tz :D | 09:40 |
rpittau | :D | 09:40 |
JayF | or well, I'm in yours or closer to it | 09:40 |
rpittau | we're a train ride away! | 09:41 |
JayF | I'll note I'm in Paris on Monday with no plans, if it's close enough and you wanna grab lunch or a chat we can, but otherwise I'd see you Weds/Thurs | 09:41 |
vicent | Hi, I hope this is the place to ask about ironic. I am exploring the multi-tenant networking on devstack. I tried both https://docs.openstack.org/networking-baremetal/latest/contributor/quickstart-multitenant.html and https://docs.openstack.org/ironic/latest/contributor/ironic-multitenant-networking.html none of those work on a clean ubuntu 22.04 VM. I guess the difference is the first uses OVS | 09:55 |
vicent | and the second OVN. I could make the first work with some manual changes and restarting services. Which one would be the recommended way though? | 09:55 |
JayF | Hey vicent o/ | 09:55 |
JayF | I'm not surprised those devstack configs are somewhat outta date, sadly | 09:56 |
JayF | I'm sure what you want to do is possible but it may take a little bit of coaxing | 09:56 |
JayF | although honestly that ngs (networking-generic-switch) devstack config looks pretty good | 09:56 |
JayF | Is your goal just to try out Ironic multitenant networking? To do development? What's the bigger goal you wanna accomplish | 09:57 |
vicent | I had to comment out the network for instances variables. 10.0.1.0/24 somehow didn't work. Then I had to restart the ir-neutronagt in the middle of the stack.sh. Then it worked | 09:58 |
vicent | I also have to remove the dnsmasq folder between instantiations. That one and the ~/.vbmc folder don't get cleaned. | 09:59 |
JayF | is it possible you have 10.0.1.0/24 on your local subnet? | 09:59 |
vicent | My goal for now it is to understand the possibilities or multitenant networking on ironic and maybe some development. | 10:00 |
JayF | Also, generally speaking, it's best to restart with a clean ubuntu when trying to re-stack a devstack setup IME. I usually keep a snapshot of everything ready before stack.sh. | 10:00 |
JayF | I suspect if you've been trying to modify config and re-stack that may explain some of your difficulty | 10:00 |
vicent | I see, I will try that | 10:01 |
vicent | No, 10.0.1.0/24 is not on my local subnet | 10:02 |
JayF | Yeah definately make sure your local LAN isn't overlapping the network setup there | 10:02 |
JayF | darn | 10:02 |
JayF | I've not done serious development on NGS in a while, or I'd give more specific help. | 10:02 |
vicent | But using the default is fine, so I will keep that. | 10:02 |
JayF | Please report back if you don't have success, we'll want to get that doc fixed up. | 10:03 |
JayF | I'll note that the Ironic (not NGS/multitenant) devstack guide is more up to date and may have useful pointers | 10:03 |
vicent | I have a conceptual question. Do we have the tenant/provider distinction on ironic? at the end both would be mapping to a physical network on the ToR switch... | 10:04 |
JayF | so cleaning_network / provisioning_network / service_network (and so forth) | 10:04 |
JayF | are the provider networks, basically | 10:04 |
JayF | where nodes are when they are not running a workload | 10:04 |
vicent | And when running a workload? | 10:05 |
JayF | then at deploy time, once the image is written, I think the flow is we power off the node; make calls to neutron (which may call NGS when configured) to change networks; boot the machine into a tenant network | 10:05 |
JayF | it's all incredibly slick when configured; but admittedly getting all the pieces in line can be tough the first time | 10:05 |
JayF | tl;dr: yes we support tenant networks | 10:05 |
vicent | I see usually those tenant networks are using vlans. Can I use other tunnel mechanisms there, for instance vxlan or geneve? | 10:06 |
vicent | assuming the TOR switch supports them | 10:07 |
JayF | I think technically we might support vxlan but not vlan | 10:07 |
JayF | it's a common patch to allow vlans through, and we're working on supporting it more officially without a neutron: https://review.opendev.org/c/openstack/ironic-specs/+/916126 | 10:08 |
JayF | I'll note we have more people familiar with NGS in American TZs, I'm usually PST but am visiting UK this week :D | 10:08 |
vicent | my devstack local.conf have Q_ML2_TENANT_NETWORK_TYPE=vlan , and that seems to be working. On the switch the tenant and provider networks looks the same. | 10:08 |
vicent | Ok, I will wait tomorrow to read the comments :) | 10:09 |
JayF | TheJulia: ^ if you have time when you're up, you might be able to say a couple sentences and help vicent get pointed in the right direction | 10:10 |
vicent | Another question. Does NGS work with multiple switches? I saw several can be configured on the ngs ml2 file, but I was wondering if neutron would know to configure the tenant and provider networks on multiple switches. | 10:12 |
JayF | Yes; and I don't know how :) | 10:14 |
JayF | sorry you got the IPA expert not the NGS expert :D | 10:14 |
JayF | ask me about cleaning nodes and you will get less i don't knows LOL | 10:14 |
JayF | something that may help you out | 10:14 |
JayF | push a patch to ngs, literally anything | 10:14 |
JayF | look at the CI logs | 10:14 |
JayF | that might give you some insight into what a working setup looks like | 10:15 |
JayF | https://review.opendev.org/c/openstack/networking-generic-switch/+/919593 has a recent result you could look at | 10:16 |
JayF | with https://zuul.opendev.org/t/openstack/build/18acda8b806e49159952bf30e8122914being most likely the best relevant example | 10:16 |
JayF | (it's multinode which means we run two devstacks and they coordinate, it can be easier to test networking that way) | 10:16 |
JayF | ah no, the logs are gone because it's too old | 10:17 |
JayF | I'll recheck it, it'll have new logs in about an hour | 10:17 |
vicent | I get an error when trying to view the build. Do I need to sign in? | 10:17 |
JayF | no, it's because it's too old, I should've clicked before copying it | 10:17 |
JayF | I approved that change, it'll run CI again, look for the comment from zuul voting "Verified+2" in about an hour and it'll have a new set of logs | 10:18 |
JayF | they expire after a few days | 10:18 |
JayF | (we set up and tear down thousands of clouds every day so we can't keep all the info forever :D) | 10:18 |
vicent | Ok, thanks for the hint JayF. I will keep looking. | 10:19 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Remove outdated information from contributor guide https://review.opendev.org/c/openstack/ironic/+/920649 | 10:19 |
JayF | thanks for the inspiration to do ^ that | 10:20 |
JayF | :D | 10:20 |
JayF | having the less-tested devstack configs on the front door with the detailed ones hidden is not ideal | 10:20 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Update front page of contributor guide https://review.opendev.org/c/openstack/ironic/+/920649 | 10:35 |
iurygregory | good morning ironic o/ | 10:48 |
opendevreview | Dmitry Tantsur proposed openstack/ironic master: [PoC] Render the redfish interop profile in the docs https://review.opendev.org/c/openstack/ironic/+/920574 | 10:50 |
JayF | o/ | 10:51 |
JayF | iurygregory: https://review.opendev.org/c/openstack/ironic/+/920365 needs one more core review to unbreak the gate | 11:26 |
JayF | if you can, it'd be helpful | 11:26 |
iurygregory | JayF, sure looking now | 11:26 |
opendevreview | Merged openstack/networking-generic-switch master: [CI] Remove not defined job from queue https://review.opendev.org/c/openstack/networking-generic-switch/+/919593 | 11:45 |
opendevreview | Kaifeng Wang proposed openstack/ironic master: Fix some typo in servicing doc https://review.opendev.org/c/openstack/ironic/+/920323 | 11:46 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Update front page of contributor guide https://review.opendev.org/c/openstack/ironic/+/920649 | 12:16 |
opendevreview | Elod Illes proposed openstack/networking-generic-switch stable/2024.1: [CI] Remove not defined job from queue https://review.opendev.org/c/openstack/networking-generic-switch/+/920655 | 12:31 |
opendevreview | Elod Illes proposed openstack/networking-generic-switch stable/2023.2: [CI] Remove not defined job from queue https://review.opendev.org/c/openstack/networking-generic-switch/+/920659 | 12:38 |
opendevreview | Elod Illes proposed openstack/networking-generic-switch stable/2023.1: [CI] Remove not defined job from queue https://review.opendev.org/c/openstack/networking-generic-switch/+/920660 | 12:39 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Update front page of contributor guide https://review.opendev.org/c/openstack/ironic/+/920649 | 12:42 |
opendevreview | Verification of a change to openstack/ironic master failed: Fix CI issues in codespell/pep8 https://review.opendev.org/c/openstack/ironic/+/920365 | 12:46 |
opendevreview | Verification of a change to openstack/ironic master failed: Fix CI issues in codespell/pep8 https://review.opendev.org/c/openstack/ironic/+/920365 | 13:02 |
TheJulia | good morning | 13:08 |
pocoyo_2468 | /who | 13:26 |
opendevreview | Merged openstack/ironic-inspector master: Fix codespell reported errors https://review.opendev.org/c/openstack/ironic-inspector/+/920399 | 13:29 |
pocoyo_2468 | /help | 13:30 |
* TheJulia raises an eyebrow | 13:30 | |
pocoyo_2468 | /msg ebilj did u receive this | 13:30 |
TheJulia | pocoyo_2468: could you be in the wrong channel? | 13:30 |
pocoyo_2468 | i am not sure | 13:31 |
TheJulia | What channel were you seeking? | 13:31 |
pocoyo_2468 | i am just trying a vscode extension that acts as an irc client | 13:32 |
pocoyo_2468 | i got to here and now i am unsure how to send direct msg to my colleague | 13:32 |
TheJulia | Interesting | 13:33 |
TheJulia | I've not heard of such an extension | 13:33 |
pocoyo_2468 | it's called vsirc | 13:33 |
TheJulia | neat, but it sounds/seems like from your messages into channel that it doesn't really grok / commands | 13:36 |
TheJulia | That or it expects those commands to be sent on like a server tab | 13:36 |
pocoyo | maybe but everything so far used / commands | 13:41 |
pocoyo_2468 | first i used: /connect irc.oftc.net | 13:41 |
pocoyo_2468 | then: /join #openstack-ironic | 13:42 |
pocoyo_2468 | and now i am here ☺️ | 13:42 |
dtantsur | curious | 13:47 |
dtantsur | TheJulia, JayF, a question on the Redfish profile versioning (tagging you two since the three of us are most likely to care) | 13:47 |
dtantsur | https://opendev.org/openstack/ironic/src/branch/master/redfish-interop-profiles/OpenStackIronicProfile.v1_0_0.json has 1.0.0 inside and its file name, and this seems to be by design | 13:47 |
dtantsur | I'm not changing/fixing it. Is the expectation that I actually leave the old file alone and create a 1_1_0? | 13:48 |
dtantsur | The old file seems to be slightly wrong in a few places, but nothing catastrophic | 13:48 |
dtantsur | I suspect the DMTF way is to freeze the old version, but I wonder how much of a heavyweight process it is for us | 13:49 |
dtantsur | And how much we care about a not-quite-correct profile still available in master | 13:49 |
dtantsur | Opinions? | 13:49 |
TheJulia | I'd just fix the file and update the versions to match latest, I'm not sure we can enforce version freezing realistically | 13:56 |
dtantsur | TheJulia: latest = 1.1.0 in this case? or you mean the version of ironic itself? | 14:02 |
TheJulia | 1.1.0 I think | 14:03 |
TheJulia | since it would be versioned separately | 14:03 |
dtantsur | ++ | 14:03 |
dtantsur | TheJulia: re https://bugs.launchpad.net/ironic/+bug/2067073: you mention iPXE incompatibility with the secure boot, but grub does not have this problem, does it? | 14:21 |
dtantsur | Not objecting to ISOs, it will be even handier for us in OpenShift, just clarifying | 14:22 |
TheJulia | For grub, you *can* load shim directly, and shim will then load the grub efi binary | 14:22 |
TheJulia | but shim is married only to grub, and stevebaker did sort of push shim folks to see if we could also support loading ipxe, it would need to be signed by the vendor's key loaded in shim and it would need to be named the same static filename as grub | 14:23 |
TheJulia | .... which is obviously far from ideal as well | 14:23 |
TheJulia | But, you can just use grub if compiled for network booting | 14:24 |
TheJulia | well, shim+grub | 14:24 |
dtantsur | yeah, that's what I meant | 14:26 |
dtantsur | I'm also wondering if the difference between an ISO and grub boot is significant enough to justify a new boot interface | 14:26 |
TheJulia | in ironic's case of configuration, boot_interface needs to be set to pxe, the associated template needs to be the grub config, the grub initial chain loader config needs to be grub.cfg | 14:26 |
dtantsur | yeah, it's probably different enough | 14:27 |
dtantsur | for ISOs, we need to support both building them and taking provided images (like for virtual media) | 14:27 |
dtantsur | quite a lot of differences | 14:27 |
TheJulia | pxe was always intended to be a "you just pxe boot" and we just mix and match, I still see people who use it with ipxe.... | 14:27 |
TheJulia | yeah | 14:27 |
dtantsur | okay, I'm +2 to the RFE | 14:28 |
TheJulia | cool cool | 14:28 |
opendevreview | Jay Faulkner proposed openstack/ironic master: Update front page of contributor guide https://review.opendev.org/c/openstack/ironic/+/920649 | 14:34 |
dtantsur | https://review.opendev.org/c/openstack/ironic/+/920574 is my WIP effort to fix and render the profile | 14:35 |
JayF | What is the /use case/ for 2067073? I am struggling to understand when you'd want ironic to send an iso vs bootloader | 14:35 |
JayF | It seems like a different-in-name/format only option; which implies to me I'm missing something | 14:36 |
TheJulia | Basically, different formatting/modeling. The use case in mind is to support direct network booting of host from UEFI firmware so they can boot with Secure Boot directly from the ISO. The only way we do that today is via the OOB URL, but there are still people (still!) who insist on IPMI, or who just want to network boot all the time (i.e. some can't support out of band BMC connectivity for virtual media. | 14:39 |
JayF | so to clarify: | 14:40 |
JayF | existing way is multi-stage, you get a loader (ipxe/grub) which gets a thing | 14:40 |
JayF | new proposed way is one stage, you get an iso which has it all baked in | 14:40 |
TheJulia | yes, and the important fact is ipxe is not signed, nor will ever be publically signed most likely | 14:41 |
JayF | I'm assuming this isn't the default for $hardware_support reasons, lowest common demominator hardware won't have it? | 14:41 |
TheJulia | AND EU entities are moving towards mandating SecureBoot in all operatin gmodes | 14:41 |
TheJulia | operating modes | 14:41 |
JayF | it doesn't sound like there's any downside to the ISO-based method as described | 14:41 |
TheJulia | firmware support being the needful | 14:41 |
JayF | so here's a question, I'm legitimately curious and it *feels* like something we should have in our docs | 14:42 |
TheJulia | It really is just another variation of httpboot, one I sort of skipped past and only did it for Redfish HttpBootUri support | 14:42 |
JayF | if you had hardware that could do * today | 14:42 |
JayF | what is the "best" for most use cases? | 14:42 |
JayF | redfish+http-iso? | 14:42 |
JayF | redfish+virtual media, actually? | 14:42 |
TheJulia | really it depends on the use case and capabilities | 14:43 |
JayF | I'm saying ignore capabilities, assume this is magic hardware which implements everything including mutually exclusive stuff, somehow | 14:43 |
TheJulia | My experience thus far is that tends to be magical hardware which doesn't exist ;) | 14:44 |
TheJulia | Mix in the operational constraints and folks quickly develop opinions | 14:44 |
TheJulia | But, I do think your raising a really valid point, what do we recommend | 14:44 |
TheJulia | and we sort of let people "figure that out on their own" | 14:44 |
JayF | another way of saying that is "BYODecoderRing | 14:44 |
JayF | lol | 14:45 |
JayF | part of why I ask is on my slide where I say, basically "use redfish hardware" at cern | 14:45 |
JayF | I didn't know if there was a point in saying "also use a better boot interface if you can, like X/Y/Z" | 14:45 |
TheJulia | not every boot interface supports every capability either | 14:46 |
TheJulia | That is something I think we've got written down in docs, but it also is not the clearest thing either | 14:47 |
JayF | the matrix is real :D | 14:47 |
JayF | honestly, we need like one of those giant wikipedia tables, where the y is "capabilities" and the x is some combo of boot and deploy interfaces | 14:47 |
TheJulia | ... I feel like we tried to head down that path at one point and someone -2'ed it | 14:48 |
TheJulia | or -1'ed it because it was not providing the 10 foot view | 14:48 |
* TheJulia sighs | 14:49 | |
JayF | sounds like something we would've done a while ago, but not today | 14:49 |
* TheJulia goes with trying to get downstream docs sorted | 14:49 | |
JayF | good luck :) I'm about to EOD and try to keep myself awake until a proper time to sleep UK time | 14:49 |
TheJulia | Another variation of sorts is if you go look at specific feature docs, we tend to also note the details and context into that at the very beginning of the section | 14:51 |
TheJulia | "oh, I want to do z" "z is only supported with interface x,y" | 14:51 |
* JayF builds a pivot table /s | 14:51 | |
JayF | speaking of docs - https://review.opendev.org/c/openstack/ironic/+/920649 is a fix that was inspired by the chat earlier | 14:53 |
JayF | just trying to move things out of the root document that are less-updated | 14:53 |
TheJulia | Thinking about the reality of mandated secure boot, perhaps it is time to deprecate boot from volume... | 15:01 |
JayF | I have no objection | 15:02 |
JayF | but we'd need to send it around some | 15:02 |
JayF | I'll add it to my "stuff you maybe shouldn't use" slide for CERN talk | 15:03 |
TheJulia | ++ | 15:03 |
JayF | and note it basically is impossible to secure boot with bfv | 15:03 |
JayF | and see if anyone squeals | 15:03 |
TheJulia | it is not impossible, but BFV requires iPXE, so there is just a non-easy path to get there | 15:03 |
TheJulia | one we can never actually test upstream | 15:03 |
TheJulia | and one where vendors just won't engage | 15:03 |
JayF | isn't there hardware-assisted bfv? | 15:04 |
TheJulia | oh, right | 15:04 |
JayF | which doesn't need ipxe? | 15:04 |
TheJulia | yeah | 15:04 |
TheJulia | you're right there | 15:04 |
JayF | when you said that, I assumed you meant non-hardware bfv | 15:04 |
JayF | but to be fair we can't test the hardware-assisted bfv anyway | 15:04 |
TheJulia | FJ and to an extent iLO had added some support for that | 15:04 |
TheJulia | yup | 15:04 |
JayF | it's in there https://usercontent.irccloud-cdn.com/file/xo0YMuMt/image.png | 15:06 |
TheJulia | I feel like "remote Volume" might be an important delienator | 15:06 |
JayF | "Network Boot from Volume" | 15:08 |
rpittau | that falls into "remote attestation" right? which is a different thing | 15:08 |
TheJulia | attestation is the act of measuring the system secure boot runtime state and then telling an attestation server if your "good" or "not" | 15:09 |
opendevreview | Merged openstack/ironic-lib master: Fix codespell reported errors https://review.opendev.org/c/openstack/ironic-lib/+/920425 | 15:10 |
opendevreview | Merged openstack/networking-baremetal master: Fix codespell reported errors https://review.opendev.org/c/openstack/networking-baremetal/+/920427 | 15:12 |
TheJulia | vicent: so the question of ovs vs ovn is sort of a "what are you seeking" from Neutron question. The reality is OVN has OVS under the hood, but it has some additional constraints, you'll generally want to head in the OVN direction though, but it has an open list of challenges with Bare Metal host in our docs. If you don't find that link, let me know I can dig it up. | 15:15 |
TheJulia | vicent: I woudl also expect the networking-baremetal docs to be far more out of date than the ironic docs, but configuration wise ironic also has a CI job which exercises some of the aspects. I'm not sure why the 10.1.0.0/24 network didn't work but if you can provide more details regarding that it might help. .vbmc is something I think can happen if you have re-stacks which are not fully cleaned up, so it oculd be your | 15:18 |
TheJulia | process of stacking and unstacking? We tend to recommend use of sushy-tools as opposed to virtual bmc, the difference with devstack is to use redfish instead of ipmi. | 15:18 |
TheJulia | vicent: regarding vxlan or geneve, not with networking-generic-switch. With later ironic's we did add OTAP config passing to OVN, but I think you have to individually configure the switch to point into the fabric | 15:19 |
TheJulia | vicent: ngs can be configured for multiple switches as far as I'm aware | 15:20 |
TheJulia | vicent: just separate configuration sections for each switch | 15:20 |
vicent | TheJulia: I understand OVN or OVS are just a way to simulate the ToR switch on devstack, but on real baremetal those are not there. Correct? | 15:25 |
TheJulia | actually not really | 15:26 |
TheJulia | So... OVS and OVN are what drives Neutron's engagement in terms of networking in general | 15:26 |
TheJulia | in Devstack, yes, basically drive OVS (even with ovn being used) to handle port attachments/detachments for simulating baremetal switchports being toggled | 15:27 |
TheJulia | OVN/OVS still get used in terms of however you tie Neutron into the network in a full deployment | 15:27 |
vicent | But on ironic with neutron, the mech driver is baremetal with NGS talking to the physical ToR | 15:28 |
TheJulia | yes | 15:29 |
TheJulia | for neutron as a networking node, ovs/ovn is the other side of a vlan trunk port so it sees the tagged networks and does things like setup routers | 15:29 |
TheJulia | and DHCP | 15:29 |
vicent | Ok, I see, and the provision-network and the other provider infra networks | 15:30 |
TheJulia | yup | 15:30 |
TheJulia | https://docs.openstack.org/ironic/latest/admin/ovn-networking.html <-- the ovn related issues/challenges I alluded to earlier | 15:31 |
opendevreview | Merged openstack/ironic master: Fix CI issues in codespell/pep8 https://review.opendev.org/c/openstack/ironic/+/920365 | 15:35 |
opendevreview | Adam Rozman proposed openstack/sushy-tools master: improve volume boot mode support https://review.opendev.org/c/openstack/sushy-tools/+/912113 | 15:38 |
samcat116 | For that note in the OVN doc on NAT, does that only apply if your provisioning/cleaning network is itself NATed? Or does that apply to any deployed network with a FIP in front of it. I assume it wouldn't normally deploy through the FIP in that case | 15:40 |
vicent | TheJulia: Thanks at lot for your answers. I think I have what I need to continue studying and trying things. JayF mentioned project mercury: https://review.opendev.org/c/openstack/ironic-specs/+/916126/6/specs/approved/mercury.rst It seems interesting. I will try to keep an eye on it. | 15:43 |
opendevreview | Dmitry Tantsur proposed openstack/ironic-inspector master: [DNM] Instrument to debug memory leaks https://review.opendev.org/c/openstack/ironic-inspector/+/920275 | 15:47 |
TheJulia | samcat116: if the cleaning/provision networks are natted | 15:47 |
TheJulia | err | 15:48 |
TheJulia | samcat116: I don't remember how that came up at this point | 15:48 |
TheJulia | vicent: thanks, trying to build initial interest | 15:49 |
opendevreview | Merged openstack/sushy-tools master: Fix codespell reported errors https://review.opendev.org/c/openstack/sushy-tools/+/920429 | 15:52 |
rpittau | good night! o/ | 15:56 |
opendevreview | Merged openstack/networking-generic-switch stable/2024.1: [CI] Remove not defined job from queue https://review.opendev.org/c/openstack/networking-generic-switch/+/920655 | 16:00 |
opendevreview | Merged openstack/networking-generic-switch master: Fix codespell reported errors https://review.opendev.org/c/openstack/networking-generic-switch/+/920428 | 16:01 |
opendevreview | Adam Rozman proposed openstack/sushy-tools master: improve volume boot mode support https://review.opendev.org/c/openstack/sushy-tools/+/912113 | 16:02 |
opendevreview | Adam Rozman proposed openstack/sushy-tools master: improve volume support in nova driver https://review.opendev.org/c/openstack/sushy-tools/+/912113 | 16:05 |
opendevreview | Adam Rozman proposed openstack/sushy-tools master: improve volume support in nova driver https://review.opendev.org/c/openstack/sushy-tools/+/912113 | 16:06 |
opendevreview | Adam Rozman proposed openstack/sushy-tools master: improve volume support in nova driver https://review.opendev.org/c/openstack/sushy-tools/+/912113 | 16:08 |
opendevreview | Merged openstack/ironic-inspector master: Fix negative lookahead assertion syntax error https://review.opendev.org/c/openstack/ironic-inspector/+/920281 | 16:24 |
opendevreview | cid proposed openstack/ironic master: wip: Provision aarch64 fake-bare-metal-vms https://review.opendev.org/c/openstack/ironic/+/915441 | 16:29 |
TheJulia | cid: wow that failed fast | 17:05 |
TheJulia | I'm sensing config wise soemthing is little misaligned there | 17:06 |
cid | TheJulia: One of the fastest :D | 17:09 |
TheJulia | failing fast is always best :) | 17:09 |
cid | I think I know what the problem is too | 17:09 |
opendevreview | Julia Kreger proposed openstack/ironic-python-agent-builder master: Block cloud-init user-data loading https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/920184 | 17:12 |
cid | TheJulia: I think the sooner, the closer home the problem might be, in this case, config. | 17:12 |
opendevreview | Julia Kreger proposed openstack/ironic-python-agent-builder master: Block glean-early.sh from running https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/920183 | 17:23 |
opendevreview | cid proposed openstack/ironic master: wip: Provision aarch64 fake-bare-metal-vms https://review.opendev.org/c/openstack/ironic/+/915441 | 17:35 |
opendevreview | Julia Kreger proposed openstack/ironic master: WIP Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 17:35 |
cid | TheJulia: If you get a free hand before EOD, I would like to learn how to implement the latest review on this: https://review.opendev.org/c/openstack/ironic/+/917229 | 17:36 |
TheJulia | hmmm | 17:40 |
TheJulia | oh right | 17:42 |
TheJulia | so my comment points to not using os to modify the host process environment, but to call execute() with env_variables={'VARIABLE_NAME': 'value'} | 17:42 |
cid | Yes, done that. | 17:43 |
TheJulia | okay, which latest review then? | 17:43 |
cid | I can't seem to be able to replace the _persist_ipmi_password with the _exec_ipmitool | 17:43 |
cid | As JayF suggested | 17:44 |
cid | I can push what I have going, but it's not passing unittest | 17:46 |
TheJulia | yeah, that would likely help | 17:48 |
TheJulia | Hmm, I forgot how complex this whole exec is | 17:59 |
TheJulia | I guess I might make _make_password_file somewhat of a noop with minimal code, and separately assemble the environment variable | 18:04 |
opendevreview | cid proposed openstack/ironic master: Flexible IPMI credential persistence method configuration https://review.opendev.org/c/openstack/ironic/+/917229 | 18:08 |
cid | Hmm,,, | 18:11 |
TheJulia | Yeah, I guess I'd do something like "yield None" in the method to create the fileand that could also be a signal to handle it as the environment variable | 18:14 |
cid | Sorry, I think you misunderstand me, or ... | 18:16 |
cid | I was referring to JayF's comment: "Unless there's a good reason it's not, we should switch this code to use _exec_ipmitool -- it's the only code that's independently managing credentials." | 18:16 |
cid | #1575 in the change. | 18:17 |
TheJulia | oh | 18:19 |
TheJulia | ahh | 18:19 |
TheJulia | it is generating the full command, which is why | 18:20 |
cid | Yea. | 18:23 |
cid | Ok, please, let me let you ... thanks! | 18:23 |
TheJulia | I think your on the right path there, just unit tests will need to be changed | 18:26 |
cid | noted | 18:35 |
opendevreview | cid proposed openstack/ironic master: Flexible IPMI credential persistence method configuration https://review.opendev.org/c/openstack/ironic/+/917229 | 18:47 |
cid | \o | 20:12 |
opendevreview | Merged openstack/ironic-python-agent-builder master: Block cloud-init user-data loading https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/920184 | 21:23 |
opendevreview | Julia Kreger proposed openstack/ironic master: WIP Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 21:49 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!