| opendevreview | Steve Baker proposed openstack/ironic master: Add ironic-novncproxy service https://review.opendev.org/c/openstack/ironic/+/939191 | 03:22 |
|---|---|---|
| opendevreview | Steve Baker proposed openstack/ironic master: Add systemd provider for console containers https://review.opendev.org/c/openstack/ironic/+/941614 | 03:22 |
| opendevreview | Steve Baker proposed openstack/ironic master: [WIP] implement drivers redfish-graphical, fake-graphical https://review.opendev.org/c/openstack/ironic/+/941615 | 03:22 |
| opendevreview | Steve Baker proposed openstack/ironic master: Add systemd provider for console containers https://review.opendev.org/c/openstack/ironic/+/941614 | 03:41 |
| opendevreview | Steve Baker proposed openstack/ironic master: [WIP] implement drivers redfish-graphical, fake-graphical https://review.opendev.org/c/openstack/ironic/+/941615 | 03:41 |
| opendevreview | Steve Baker proposed openstack/ironic master: Add systemd provider for console containers https://review.opendev.org/c/openstack/ironic/+/941614 | 03:52 |
| opendevreview | Steve Baker proposed openstack/ironic master: [WIP] implement drivers redfish-graphical, fake-graphical https://review.opendev.org/c/openstack/ironic/+/941615 | 03:52 |
| opendevreview | Vasyl Saienko proposed openstack/networking-generic-switch master: Add vlan aware VMs support https://review.opendev.org/c/openstack/networking-generic-switch/+/928490 | 06:43 |
| rpittau | good morning ironic! o/ | 07:45 |
| rpittau | JayF: I think we're good for https://review.opendev.org/c/openstack/governance/+/939278 ? | 10:08 |
| opendevreview | Verification of a change to openstack/ironic master failed: trivial: de-distro dnsmasq version check https://review.opendev.org/c/openstack/ironic/+/941709 | 12:21 |
| opendevreview | Verification of a change to openstack/ironic master failed: trivial: lock dnsmasq check to ubuntu https://review.opendev.org/c/openstack/ironic/+/941710 | 12:21 |
| opendevreview | yatin proposed openstack/ironic stable/2024.1: Pass agent token to get command results https://review.opendev.org/c/openstack/ironic/+/941912 | 12:50 |
| adam-metal3 | JayF: we might have some issues with thi https://review.opendev.org/c/openstack/ironic-python-agent/+/941539 in Metal3 CI, folks have pinged you on slack but is this feature intended to be backward compatible? | 13:17 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add tests to validate trunks https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 | 13:33 |
| opendevreview | yatin proposed openstack/ironic stable/2023.2: Pass agent token to get command results https://review.opendev.org/c/openstack/ironic/+/941917 | 13:35 |
| dtantsur | adam-metal3: oh.. it should not have landed like this | 13:35 |
| dtantsur | it's very-very much not backward compatible, and there is not a single word about it in the release note | 13:36 |
| dtantsur | in fact, I highly suspect that the release note has not been reviewed at all | 13:36 |
| ykarel | dtantsur, seems you talking about https://review.opendev.org/c/openstack/ironic-python-agent/+/941539 | 13:37 |
| ykarel | will that be reverted? | 13:37 |
| ykarel | asking because that broke jobs running with stable/2024.1 and before | 13:37 |
| dtantsur | ykarel: are you also affected? I feel like it should be reverted and added with a graceful compatibility period | 13:37 |
| ykarel | dtantsur, actually in neutron periodic we have ironic job failing due to this as missing backport in ironic | 13:38 |
| ykarel | with backport https://review.opendev.org/q/Iad0d8086e1e79b59da3f9940b83128a3b34d4787 it passing | 13:38 |
| ykarel | so either revert or merge of backport will work for us to clear this | 13:39 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent master: Revert "Add token validation to GET command endpoints" https://review.opendev.org/c/openstack/ironic-python-agent/+/941919 | 13:40 |
| dtantsur | ykarel, JayF, TheJulia, cid, sorry but ^^^ | 13:40 |
| ykarel | dtantsur, thx for taking care of that, will abandon the proposed backport | 13:41 |
| adam-metal3 | dtantsur: thanks for the confirmation | 13:41 |
| dtantsur | ykarel: I do think we should backport the Ironic change | 13:42 |
| ykarel | dtantsur, ok then can keep that up for review | 13:43 |
| dtantsur | one day, we will force the token validation, it's good to have the token sent even on earlier versions since people do mix-and-match | 13:43 |
| ykarel | ack got it | 13:43 |
| dtantsur | ykarel: as an aside, I'm quite curious how the neutron job ended up using master IPA with an old Ironic | 13:43 |
| ykarel | dtantsur, it's an ironic job only we just run in neutron periodic https://zuul.openstack.org/job/ironic-tempest-ipa-wholedisk-bios-agent_ipmitool-tinyipa | 13:44 |
| ykarel | if that's not expected combination ^ job should be fixed, but considering IPA branchless i assumed it to work with couple of stable branches | 13:45 |
| dtantsur | ykarel: IPA is not branchless though | 13:46 |
| dtantsur | I'm curious if we ended up with an invalid configuration somehow | 13:46 |
| ykarel | ohh somehow i recalled it wrongly then | 13:46 |
| dtantsur | https://opendev.org/openstack/ironic/src/branch/master/devstack/lib/ironic#L334 | 13:47 |
| dtantsur | which... was not updated for 2024.1 \o/ https://opendev.org/openstack/ironic/src/branch/stable/2024.1/devstack/lib/ironic#L327 | 13:47 |
| ykarel | then it's surely misconfig, thx for pointing it | 13:47 |
| * dtantsur is getting reaaaaaally angry today | 13:47 | |
| dtantsur | starting with 2024.2, we at least default to TARGET_BRANCH, but it was not the case in 2024.1 | 13:48 |
| ykarel | ^ good | 13:48 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic stable/2024.1: ci: automatically set the branch for IPA to match TARGET_BRANCH https://review.opendev.org/c/openstack/ironic/+/941920 | 13:49 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic stable/2023.2: ci: automatically set the branch for IPA to match TARGET_BRANCH https://review.opendev.org/c/openstack/ironic/+/941921 | 13:49 |
| TheJulia | good morning | 14:12 |
| JayF | dtantsur: I guess I'm confused because we didn't give a transition period for the image security bug fix | 14:39 |
| JayF | dtantsur: You're right about the release note, but it's not clear to me that we always allow this transition. Perhaps that was just a mistake with the image migration stuff? Or at least we had a flag there to disable it? | 14:40 |
| JayF | **mitigation | 14:40 |
| JayF | Yeah, that's probably the right shape isn't it, configuration as to whether or not it has to be authenticated... | 14:41 |
| dtantsur | The image mitigation was sorta special, but also it was not supposed to affect the normal path for most (hopefully all) consumers | 14:44 |
| dtantsur | We might, as a community, to decide to land a breaking change, but we need to be really careful with the messaging. | 14:44 |
| dtantsur | Or we can tolerate a missing token for this release but make it a fatal error early in the next one | 14:44 |
| dtantsur | But I don't want the image validation to become a precedent for a quick-and-dirty approach. It was very special in many ways. | 14:46 |
| JayF | I don't disagree with you necessarily, but I think a more formal ironic/ IPA compatibility policy may be in order. Because I think there's not consensus currently on how careful we have to be | 14:56 |
| JayF | We certainly don't test any combination other than latest ironic and latest IPA and released ironic and released IPA | 14:57 |
| JayF | (at least intentionally) | 14:57 |
| TheJulia | A thought is to enable the agent to recognize the message and lockout further command calls | 14:57 |
| TheJulia | the first command is actually to get steps | 14:57 |
| TheJulia | that data gets retrieved | 14:57 |
| TheJulia | err, not command calls, but command call results | 14:58 |
| JayF | Yeah, that's what I meant when I said it would be configurable, that the conductor would turn it on during lookup | 14:59 |
| JayF | And then at some point in the future we could default it to on and remove the setting | 14:59 |
| TheJulia | Well, the agent learning and auto-restricting is likely cleanest because it could be backported without issue, wouldn't require parameter passing | 15:00 |
| TheJulia | It... could break if someone is deploying in a mixed upgrade state though, but they are supposed to upgrade the conductors together | 15:00 |
| TheJulia | Anyway! | 15:00 |
| TheJulia | #startmeeting ironic | 15:01 |
| opendevmeet | Meeting started Mon Feb 17 15:01:03 2025 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:01 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:01 |
| opendevmeet | The meeting name has been set to 'ironic' | 15:01 |
| TheJulia | o/ | 15:01 |
| masghar | o/ | 15:01 |
| rpittau | you got me | 15:01 |
| rpittau | o/ | 15:01 |
| TheJulia | #chair rpittau | 15:01 |
| opendevmeet | Current chairs: TheJulia rpittau | 15:01 |
| rpittau | Hello everyone! | 15:01 |
| rpittau | Welcome to our weekly meeting! | 15:01 |
| rpittau | I'm fighting with the wiki page | 15:01 |
| TheJulia | I believe the rules say you must always struggle with the wiki page right before the meeting ;) | 15:02 |
| cid | o/ | 15:02 |
| rpittau | meeting agenda should be available here though: | 15:02 |
| rpittau | https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_February_17.2C_2025 | 15:02 |
| rpittau | seem to work now :) | 15:02 |
| rpittau | #topic Announcements/Reminders | 15:03 |
| rpittau | Standing reminder to review patches tagged ironic-week-prio and to hashtag any patches ready for review with ironic-week-prio: | 15:03 |
| rpittau | #link https://tinyurl.com/ironic-weekly-prio-dash | 15:03 |
| rpittau | we're looking good there | 15:03 |
| rpittau | 2025.1 Epoxy Release Schedule | 15:05 |
| rpittau | #link https://releases.openstack.org/epoxy/schedule.html | 15:05 |
| rpittau | we're at R-6 | 15:05 |
| rpittau | this week we're releasing metalsmith and sushy | 15:05 |
| TheJulia | Does sushy have anything outstanding? | 15:06 |
| TheJulia | And... why are we continuing to release Metalsmith? | 15:06 |
| rpittau | I've blocked the sushy release for the time benig because of the migration of sushy-oem-idrac | 15:06 |
| rpittau | well metalsmith is not deprecated, isn't it? | 15:06 |
| dtantsur | TheJulia: we cannot just stop releasing something | 15:06 |
| TheJulia | Its a zombie project at this point | 15:06 |
| rpittau | until we deprecate/retire it we have to release it | 15:07 |
| TheJulia | We need to | 15:07 |
| JayF | If the team comes to a consensus we should do that, I'll run the paperwork for it since I've already got it done for ir-lib | 15:07 |
| TheJulia | But I'm not going to start a fight or argue anymore about it | 15:07 |
| JayF | I'm not sure if it's too late in this cycle or not | 15:07 |
| TheJulia | I've got big enough stuff to worry about | 15:07 |
| cardoe | I can switch back to doing sushy-oem-idrac work. | 15:07 |
| rpittau | JayF: it's probably too late, the release is happening now :) | 15:07 |
| JayF | Yeah I think that's actually the real problem is that no one thinks about metalsmith at all until someone asks us to release it | 15:08 |
| cardoe | There's some issues in the tests for sushy that installing another oem extension breaks the tests. I believe the actual code of sushy is correct. | 15:08 |
| dtantsur | The price of releasing it is quite low | 15:08 |
| rpittau | or until there is a CI blocker | 15:08 |
| cardoe | So let's deprecate metalsmith unless someone else steps up to take over maintaining it? | 15:08 |
| rpittau | cardoe: that was one of the topics to discuss, if you have time and we want to have that done for epoxy it needs to happen this week | 15:09 |
| rpittau | sushy-oem-drac nto sushy ^ | 15:09 |
| cardoe | okay I'll make it happen if I get the reviews. | 15:09 |
| rpittau | (my keyboard is skipping the Is very often...) | 15:09 |
| rpittau | cardoe: happy to review :) | 15:10 |
| rpittau | BTW next week s feature freeze and final release for clients | 15:10 |
| rpittau | as well as requrements freeze | 15:10 |
| cardoe | The patch I tossed together seems to work fine. It's literally just the tests for looking up and loading oem extensions and testing that the right one was loaded. The tests monkey patch in an extension into the steveadore list and expect the fake extension to be at [1] in the list. | 15:10 |
| cardoe | afaik the real list now is [0]=conoco, [1]=dell, [2]=fake | 15:11 |
| dtantsur | Are we sure installing patched sushy together with sushy-oem-idrac yields a predictable result? | 15:11 |
| cardoe | It's honestly no different than separate today. | 15:13 |
| rpittau | still, we need to make sure that sushy-oem-idrac is not installed anymore | 15:14 |
| cardoe | Yeah need to make a conflict in the versioning | 15:14 |
| cardoe | It's more of that machinery needing to happen for any tests. | 15:14 |
| rpittau | ok, let's see if we can make it | 15:17 |
| rpittau | please consider that I'm off this Friday, in case reviews are needed | 15:17 |
| rpittau | the sushy release patch is https://review.opendev.org/c/openstack/releases/+/941886 | 15:17 |
| rpittau | next reminder is for the vPTG for Flamingo that will take place place April 7-11, 2025 | 15:18 |
| rpittau | I'll have the etherpad up for topcs this week | 15:19 |
| rpittau | (need to change keyboard) | 15:19 |
| TheJulia | Excellent | 15:19 |
| TheJulia | :( | 15:19 |
| TheJulia | :( in response to need to change keyboard | 15:19 |
| rpittau | :D | 15:19 |
| rpittau | anything else to announce/remind? | 15:20 |
| rpittau | ok moving on | 15:20 |
| rpittau | #topic Discussion topics | 15:20 |
| rpittau | #topic change to DPL model for ironic? | 15:21 |
| rpittau | I guess we reached some kind of consensus there | 15:21 |
| rpittau | I've added the list of the proposed liaisons to the agenda | 15:21 |
| TheJulia | Seems like it | 15:21 |
| rpittau | if we're ok with that we can move forward with the bureaucratic stuff I guess | 15:21 |
| TheJulia | No objection from me | 15:22 |
| dtantsur | ++ | 15:22 |
| rpittau | alright, if there's no objection I will start with the changes this week or early next week | 15:23 |
| TheJulia | Where ever did the realization that this had to be put in place *before* the election end up? | 15:24 |
| TheJulia | s/election/ptl nomination/ | 15:24 |
| JayF | That's the way it's always worked. I honestly think it's kind of weird that DPL can happen without an election in general, but I guess TC is the election that is the backstop for that | 15:25 |
| JayF | It is only a recent thing that DPL has been used by active projects. | 15:25 |
| JayF | (more active,I guess) | 15:25 |
| cardoe | So let's just pick someone to be PTL for the "official" aspect. We've agreed on DPL. Once the election hump happens. I'll convert us to DPL. | 15:26 |
| cardoe | I'll do that TC liaison thing and just make it happen. :) | 15:26 |
| rpittau | I'm PTL until the end of the next elections, so I guess we're fine? :) | 15:26 |
| JayF | On the mailing list, a plan was laid out of going through leaderless process to become DPL. If someone tries to run for PTL out of the blue, I've been watching and will post my own candidacy just to convert to DPL. | 15:26 |
| JayF | I think that'll be easier than having people go through the motions of putting metadata in about a PTL just to replace it with metadata about a DPL | 15:27 |
| TheJulia | sounds like a plan | 15:28 |
| rpittau | great! :) | 15:28 |
| rpittau | ok, moving on | 15:29 |
| rpittau | we talked about sushy-oem-idrac | 15:29 |
| rpittau | #topic ironic-lib deprecation | 15:30 |
| rpittau | similar situation | 15:30 |
| rpittau | JayF: I think we're good | 15:30 |
| rpittau | #link https://review.opendev.org/q/topic:%22ironic-lib-deprecation%22 | 15:30 |
| JayF | Yep. I'll take a look at making sure the next patches are in a decent state and pointing the TC at them because I think governance is next. | 15:31 |
| JayF | From technical perspective though, ironic lib is dead | 15:31 |
| rpittau | long live ironic-lib! | 15:31 |
| rpittau | thanks JayF for your effort :) | 15:31 |
| rpittau | anything else we have to discuss today? | 15:32 |
| rpittau | okey dokey! | 15:33 |
| rpittau | thanks everyone! | 15:33 |
| rpittau | #endmeeting | 15:33 |
| opendevmeet | Meeting ended Mon Feb 17 15:33:51 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:33 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-17-15.01.html | 15:33 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-17-15.01.txt | 15:33 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-17-15.01.log.html | 15:33 |
| cardoe | I'm also very much trying to landing the redfish running hooks this cycle. | 15:34 |
| cardoe | I've got 2 more patches in the series that are generic that I believe are ready to go. | 15:35 |
| JayF | Fyi, if you need something specifically for me today, please speak up early as I might try to take a half day of PTO since my wife has the day off. | 15:35 |
| cardoe | https://review.opendev.org/c/openstack/ironic/+/940332 and https://review.opendev.org/c/openstack/ironic/+/940333 | 15:35 |
| cardoe | That should make everything super duper generic. | 15:35 |
| cardoe | I was thinking of landing some basic "shape of inspection data" validation before I turned on redfish so that we'd know before hooks blow up. | 15:36 |
| TheJulia | do we have consensus on forward path for ipa patch, or not? | 15:37 |
| TheJulia | w/r/t get command results response authentication | 15:37 |
| cardoe | But the reason I mention those two patches is that the first one fixes an actual bug in hooks execution in Ironic which causes an uncaught exception if there's a bug in a preprocess() hook. Ironic attempts to log the failure but the logging line is wrong inside the Ironic code base. | 15:37 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: allow multiple inspection interfaces to load hooks https://review.opendev.org/c/openstack/ironic/+/940333 | 15:38 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: allow running inspection hooks on redfish interface https://review.opendev.org/c/openstack/ironic/+/933066 | 15:38 |
| cardoe | bah. I realized just now I needed to drop the release note on patch #2. | 15:39 |
| rpittau | proposal to change ironic to DPL is submitted https://review.opendev.org/c/openstack/governance/+/941935 | 15:48 |
| JayF | oh, rpittau, I get what you mean now | 15:56 |
| JayF | if new PTL can make it DPL, you can now | 15:56 |
| rpittau | yep :) | 15:56 |
| JayF | it seems so obvious after the fact | 15:56 |
| JayF | Iury (not in channel), dtantsur, and cardoe need to +1 that to indicate they are willing to be liasons | 15:57 |
| dtantsur | I think Iury is back next week, can we use his email to the ML as a consent proof? | 15:58 |
| TheJulia | Seems reasonable to attempt | 15:58 |
| rpittau | we can give it a try | 15:59 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add tests to validate trunks https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 | 16:28 |
| opendevreview | Satoshi Shirosaka proposed openstack/ironic master: Create Support Resources Page https://review.opendev.org/c/openstack/ironic/+/941293 | 16:34 |
| rpittau | good night! o/ | 17:04 |
| TheJulia | I've changed my vote on https://review.opendev.org/c/openstack/ironic/+/699953 because it looks like we have something else going on and the current approach ends up hanging things as well. | 18:11 |
| JayF | yeah, that's an insightful comment, I already looked and am noodling on it | 18:21 |
| JayF | basically thinking if we wouldn't fail on binding failure, and if we aren't a smart nic ... continue to not to wait | 18:21 |
| TheJulia | Thinking about it | 18:21 |
| JayF | either that or a configurable timeout | 18:21 |
| TheJulia | Regardless, 20 minutes is also just unacceptably blocking when it is disabled | 18:21 |
| TheJulia | I'm super surprised that it got triggered and became blocking | 18:22 |
| JayF | yeah the situation makes sense given the code, and wasn't a case I was considering | 18:23 |
| JayF | and I suspect john never runs without networking-baremetal | 18:23 |
| TheJulia | Granted, running without is super weird | 18:25 |
| TheJulia | well, for neutron integrated | 18:25 |
| JayF | I've known many places that did, but all that experience was when ironic+networking was less mature | 18:28 |
| dking | Has anybody here had any trouble with Supermicro boxes recently? Their latest BMC firmware seems to have trouble with ipmitool -I lanplus. | 18:33 |
| opendevreview | Merged openstack/ironic-python-agent master: Revert "Add token validation to GET command endpoints" https://review.opendev.org/c/openstack/ironic-python-agent/+/941919 | 18:35 |
| dking | Well, with the latest ipmitool version, though. ipmitool 1.8.18 doesn't seem to have the issue. | 18:37 |
| JayF | dking: https://github.com/ipmitool/ipmitool/blob/master/ChangeLog#L22 looks suspicious | 18:37 |
| JayF | dking: https://github.com/ipmitool/ipmitool/commit/4d25a93f4922885c3d557bf4f9a83968ad2edee9 should shed some light | 18:38 |
| TheJulia | Yeah, cipher suite being wrong/incompatible results in non-responses | 18:41 |
| dking | Yeah, that's what I was suspecting. I opened a Supermicro ticket, but their response was asking me to use a different IPMI tool and sent me a beta version of the firmware which still gives the error. So, I imagine the next step is to confirm what cipher they are using and show them the specs that say its wrong. | 18:46 |
| opendevreview | Merged openstack/ironic stable/2024.1: Pass agent token to get command results https://review.opendev.org/c/openstack/ironic/+/941912 | 18:58 |
| dking | I wonder if those updates were related to the cipher 0 attacks? It would seem odd, though, to update the tool rather than the firmware. | 19:01 |
| JayF | ipmitool is pretty dead as a project iirc | 19:01 |
| JayF | that linked change is from 2022 | 19:01 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Trivial: Enable disabling tftp setup https://review.opendev.org/c/openstack/ironic/+/941742 | 19:07 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Fix devstack plugin for centos https://review.opendev.org/c/openstack/ironic/+/941720 | 19:13 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Trivial: Enable disabling tftp setup https://review.opendev.org/c/openstack/ironic/+/941742 | 19:13 |
| vsaienko | JayF, cardoe please add to your review queue, implementation of vlan aware VMS https://review.opendev.org/c/openstack/networking-generic-switch/+/928490. And related test https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 which passed on n-g-s patch. Also 1liner which enables trunk extension is needed https://review.opendev.org/c/openstack/ironic/+/941023 | 19:21 |
| opendevreview | Verification of a change to openstack/ironic master failed: trivial: de-distro dnsmasq version check https://review.opendev.org/c/openstack/ironic/+/941709 | 19:28 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add retries while waiting for SSH on server https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/942009 | 20:19 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add retries while waiting for SSH on server https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/942009 | 20:22 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add tests to validate trunks https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 | 20:23 |
| cardoe | Will do. | 20:27 |
| cardoe | I'm so deep inside of this ESXi hell scape. | 20:27 |
| cardoe | The installer really really wants its stuff loaded as multiboot | 20:27 |
| cardoe | Like Xen actually. But the Anaconda installer is quite different from multiboot unfortunately. | 20:28 |
| dking | JayF: The command is what's issued by Ironic. Are you thinking that we should be using a different tool? | 20:34 |
| TheJulia | cardoe: like different template? | 20:35 |
| JayF | dking: I'm thinking that knobs already exist to adjust the cipher suite and stuff in that command, but I'm not 100% sure as I haven't looked at the ipmi driver in approximately three centuries /s | 20:38 |
| TheJulia | Is 2018 three centuries? | 20:47 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent master: Lockout agent command results if a token is received https://review.opendev.org/c/openstack/ironic-python-agent/+/942010 | 20:51 |
| TheJulia | JayF: ^ I gave it a spin | 20:51 |
| TheJulia | locks out as I proposed earlier today | 20:52 |
| cardoe | TheJulia: sorry I was tabbed out.. | 21:10 |
| cardoe | So essentially the multiboot spec sets up a pile of modules in memory and the first one is where you jump into | 21:11 |
| cardoe | TheJulia: https://etherpad.opendev.org/p/cardoe-boot-esxi | 21:12 |
| cardoe | They use mboot.c32 from pxelinux which supports something similar as far as setup goes. | 21:15 |
| cardoe | It'll read from pxelinux.0/01-<MAC> which Ironic populates with it's own boot data. | 21:15 |
| cardoe | If it's EFI then it reads 01-<MAC>/boot.cfg | 21:16 |
| TheJulia | That sort of makes sense, that is all dhcp related to config and initial loading path | 21:16 |
| TheJulia | so you could kind of template it | 21:17 |
| cardoe | Yep. The part that I'm also fighting in here is that nova is booting this as well. | 21:18 |
| cardoe | I tried to set the deploy interface to anaconda and let nova kick the box. | 21:19 |
| cardoe | And it fails.. the error message sucks. | 21:19 |
| TheJulia | what fails where? | 21:19 |
| cardoe | So booting with a "openstack server create" fails with "Failure prepping block device." | 21:20 |
| TheJulia | best bet, ironic fails somewhere | 21:21 |
| TheJulia | I'd check ironic's logs and internally whatever is causing nova.virt.ironic to explode is likely not far behind | 21:21 |
| cardoe | Cannot validate image information for node 90b75aae-bd7b-4ad8-98f0-230968738d2c because one or more parameters are missing from its instance_info and insufficient information is present to boot from a remote volume. Missing are: ['image_source', 'kernel', 'ramdisk'] | 21:22 |
| TheJulia | do you have a kernel and ramdisk parameter on a "glance image" record for any assets required for this? | 21:23 |
| cardoe | I do | 21:23 |
| cardoe | look at the etherpad I pasted it based on the docs. | 21:24 |
| TheJulia | umm... is _id right? | 21:26 |
| cardoe | https://docs.openstack.org/ironic/latest/admin/anaconda-deploy-interface.html#configuring-the-os-image-in-glance | 21:26 |
| JayF | cid: fyi have a look at https://review.opendev.org/c/openstack/ironic-python-agent/+/942010 | 21:27 |
| JayF | TheJulia: ^ I believe CID was working on a version of that too | 21:27 |
| cardoe | The other idea is I coax nova to boot an ISO temporarily | 21:31 |
| TheJulia | okay, that is weird | 21:33 |
| TheJulia | I'd almost add some logging to see exactly what is getting set | 21:33 |
| TheJulia | oh | 21:33 |
| TheJulia | hold up | 21:33 |
| TheJulia | err, yeah | 21:33 |
| TheJulia | you should have image_source, kernel, and ramdisk based upon all your inputs | 21:33 |
| cardoe | That's what I thought. My only guess is that nova is calling validate_node() before that all gets set. | 21:35 |
| cardoe | From the error it looks like I'm going https://opendev.org/openstack/ironic/src/commit/3603e60c328cd9dd522b0080adacc86a1ea396fc/ironic/drivers/modules/deploy_utils.py#L613 | 21:39 |
| cardoe | I'd argue that nova should really set "image_source" to "glance://{uuid}" | 21:40 |
| cardoe | Cause image_source is definitely being set to that glance UUID | 21:43 |
| TheJulia | yes, likely is, however you should still have the properties | 21:45 |
| TheJulia | I'd honestly add extra logging and see what happens | 21:45 |
| cardoe | The local_link_connection is required for 'neutron' network interface and is not present in the nodes 90b75aae-bd7b-4ad8-98f0-230968738d2c port 4bf396cd-b862-46ec-9a79-9395667cc302 | 21:50 |
| cardoe | I don't necessarily believe that's causing the failure though. | 21:51 |
| cardoe | Cause it doesn't with direct. | 21:51 |
| cardoe | But it appears I cannot win here... Cause neutron won't allow me to have a local_link_connection if my network is of type "vxlan" | 21:53 |
| cardoe | But this could just be me shooting myself in the foot with tinkering with too many things at once | 21:56 |
| jamesdenton | is a custom type driver exempt from that validation? | 22:04 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent master: Remove pre-victoria cycle agent token transition upgrade support https://review.opendev.org/c/openstack/ironic-python-agent/+/941565 | 22:07 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent master: Remove agent_token_required upgrade knob https://review.opendev.org/c/openstack/ironic-python-agent/+/941566 | 22:07 |
| TheJulia | ... ugh, maybe time for a baremetal:vxlan type?! | 22:08 |
| cardoe | Well I'd like it if they let me have a local_link_connection on vxlan | 22:08 |
| TheJulia | cardoe: neutron patch time? | 22:08 |
| TheJulia | makes a ton of sense to have it I think | 22:08 |
| cardoe | The response I got back was "why it's an overlay network" | 22:09 |
| TheJulia | jamesdenton: so, I think the validation your referring to gets called regardless of, but that doesn't mean it make sense | 22:09 |
| cardoe | Except I can have more than one. | 22:09 |
| cardoe | And you could argue that one can by the overlay and one can be the underlay. | 22:09 |
| TheJulia | They don't understand vxlan in the context your wanting to use it | 22:10 |
| TheJulia | underlay is just hidden detail | 22:10 |
| jamesdenton | *thumbs up* | 22:10 |
| TheJulia | (except, with baremetal, we need local_link_information to tie it together...) | 22:10 |
| TheJulia | JayF: Updated https://review.opendev.org/c/openstack/ironic-python-agent/+/941566 but I'm not entirely sure. Time will tell I guess :) | 22:14 |
| cardoe | Well I floated out that I'd have a magical mechanism that created VLAN segments on a VXLAN network | 22:16 |
| cardoe | That the local_link_connection could be the leaf switch name in that case | 22:16 |
| opendevreview | Merged openstack/ironic master: trivial: de-distro dnsmasq version check https://review.opendev.org/c/openstack/ironic/+/941709 | 22:19 |
| opendevreview | Steve Baker proposed openstack/ironic master: Add ironic-novncproxy service https://review.opendev.org/c/openstack/ironic/+/939191 | 22:59 |
| opendevreview | Steve Baker proposed openstack/ironic master: Add systemd provider for console containers https://review.opendev.org/c/openstack/ironic/+/941614 | 22:59 |
| opendevreview | Steve Baker proposed openstack/ironic master: Implement drivers redfish-graphical, fake-graphical https://review.opendev.org/c/openstack/ironic/+/941615 | 22:59 |
| opendevreview | Steve Baker proposed openstack/ironic master: [WIP] add vnc-container image build https://review.opendev.org/c/openstack/ironic/+/942017 | 22:59 |
| cardoe | oh that was a pointless squirrel.... yes there's some extra ports on that box that don't have local_link_connection set because those aren't ports I wanna use. | 23:01 |
| cardoe | jamesdenton | 23:11 |
| cardoe | jamesdenton: so yeah thinking about making dynamic segments | 23:11 |
| jamesdenton | elaborate? | 23:14 |
| jamesdenton | in the same vein as today's network_vlan_ranges? | 23:15 |
| cardoe | yeah we can create vlan segment network ranges that are the same names as the leaf switches | 23:21 |
| cardoe | https://review.opendev.org/c/openstack/nova/+/942019 so that show fix being able to see the error message in the logs. | 23:22 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!