| opendevreview | Steve Baker proposed openstack/bifrost master: Support PXE network boot with grub https://review.opendev.org/c/openstack/bifrost/+/807220 | 02:06 |
|---|---|---|
| opendevreview | Steve Baker proposed openstack/bifrost master: Support PXE network boot with grub https://review.opendev.org/c/openstack/bifrost/+/807220 | 05:16 |
| arne_wiebalck | Good morning stevebaker[m] and Ironic! | 06:18 |
| stevebaker[m] | Good evening! | 06:28 |
| arne_wiebalck | stevebaker[m]: thanks for the quick editing of the SIG video! | 06:30 |
| stevebaker[m] | No problem:) | 06:36 |
| opendevreview | Dmitry Tantsur proposed openstack/bifrost stable/yoga: Prevent the enroll/deploy commands from running without venv https://review.opendev.org/c/openstack/bifrost/+/841876 | 10:54 |
| iurygregory | good morning Ironic | 11:05 |
| dtantsur | morning iurygregory | 11:22 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent master: Revert "The Python 3.6 and Python 3.7Support has been dropped since yaga" https://review.opendev.org/c/openstack/ironic-python-agent/+/841879 | 11:24 |
| dtantsur | iurygregory: FYI | 11:24 |
| dtantsur | ^^^ | 11:24 |
| dtantsur | also rpittau ^^^ | 11:25 |
| dtantsur | we need some multi-stage changes across ironic and IPA to fully switch to stream 9 | 11:25 |
| dtantsur | I don't want to block the CI while we're doing it | 11:25 |
| iurygregory | ++ | 11:26 |
| iurygregory | understood | 11:26 |
| opendevreview | Merged openstack/ironic-python-agent master: Revert "The Python 3.6 and Python 3.7Support has been dropped since yaga" https://review.opendev.org/c/openstack/ironic-python-agent/+/841879 | 11:48 |
| iurygregory | dtantsur, we should be good to go with some fixes now =) | 12:07 |
| dtantsur | yep. unfortunately, I rechecked some changes before | 12:33 |
| iurygregory | ouch | 12:35 |
| iurygregory | email about the group dinner in Berlin -> sent | 13:00 |
| TheJulia | good morning | 13:16 |
| dtantsur | morning TheJulia | 13:21 |
| TheJulia | it occurs to me we should have a "bios settings have changed" clean step which aborts things, and another which automatically enforces it | 13:23 |
| TheJulia | well, it being bios settings | 13:23 |
| dtantsur | not sure I'm following | 13:27 |
| dtantsur | (side note: I increasingly regret calling it "BIOS") | 13:28 |
| TheJulia | we should like also begin to port it to something else since we rarely touch that table afaik | 13:28 |
| TheJulia | so we could carry migration logic with minimal impact | 13:28 |
| TheJulia | anyway, thinking something that would help raise alarm bells or reset a prior tenant's actions | 13:29 |
| * TheJulia wonders why we don't sign metadata... | 13:55 | |
| iurygregory | good morning TheJulia =) | 14:24 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add OpenConfig classes for iface vlan plugging https://review.opendev.org/c/openstack/networking-baremetal/+/835157 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add OpenConfig classes for switch vlans https://review.opendev.org/c/openstack/networking-baremetal/+/835158 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add OpenConfig classes for network-instance https://review.opendev.org/c/openstack/networking-baremetal/+/835159 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add OpenConfig classes for interface aggregate https://review.opendev.org/c/openstack/networking-baremetal/+/835160 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add OpenConfig classes for LACP https://review.opendev.org/c/openstack/networking-baremetal/+/835161 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Device management driver iface https://review.opendev.org/c/openstack/networking-baremetal/+/835170 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add netconf-openconfig device driver https://review.opendev.org/c/openstack/networking-baremetal/+/835324 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add LACP support to Netconf OpenConfig driver https://review.opendev.org/c/openstack/networking-baremetal/+/835378 | 14:27 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Add support for pre-configured link aggregates https://review.opendev.org/c/openstack/networking-baremetal/+/840533 | 14:27 |
| ayoung | TheJulia, probably because in order to validate the signature we would have to get a valid certificate into the image. Chikcen and Egg. I'mm guessing | 14:58 |
| dtantsur | ayoung: I'm a bit out of context, but yes. We have an ability to do that for virtual media, for PXE it does not make too much sense. | 15:00 |
| iurygregory | #startmeeting ironic | 15:01 |
| opendevmeet | Meeting started Mon May 16 15:01:10 2022 UTC and is due to finish in 60 minutes. The chair is iurygregory. 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 |
| iurygregory | #chair TheJulia | 15:01 |
| opendevmeet | Current chairs: TheJulia iurygregory | 15:01 |
| dtantsur | o/ | 15:01 |
| erbarr | o/ | 15:01 |
| iurygregory | o/ | 15:01 |
| matfechner | o/ | 15:01 |
| stendulker | o/ | 15:01 |
| kamlesh6808c | o/ | 15:01 |
| rpioso | \o | 15:01 |
| rloo | o/ | 15:01 |
| arne_wiebalck | o/ | 15:02 |
| iurygregory | I will start running the meeting (I think TheJulia is busy) =) | 15:02 |
| iurygregory | Hello everyone, welcome to our weekly meeting o/ | 15:02 |
| iurygregory | you can find the agenda for this meeting in our wiki | 15:03 |
| iurygregory | #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting | 15:03 |
| iurygregory | #topic Announcements / Reminder | 15:03 |
| iurygregory | #info Please provide feedback to the Zed Themes patch | 15:03 |
| iurygregory | #link https://review.opendev.org/c/openstack/ironic-specs/+/841061 | 15:03 |
| iurygregory | #info Dinner in Berlin during the OIS | 15:04 |
| iurygregory | #link http://lists.openstack.org/pipermail/openstack-discuss/2022-May/028548.html | 15:04 |
| iurygregory | I've sent an email to the list with the details, so if you are going to the OIS and are interested check the info o/ | 15:05 |
| iurygregory | does anyone have anything else to add for announcements/reminder? | 15:05 |
| TheJulia | o/ | 15:05 |
| TheJulia | sorry | 15:05 |
| TheJulia | Got totally distracted | 15:05 |
| TheJulia | one of those "perfection is the enemy of good" discussions | 15:06 |
| hjensas | o/ | 15:06 |
| TheJulia | I take that as no announcements are desired | 15:06 |
| TheJulia | There were no action items from our meeting last week, so we can skip ahead to status reports | 15:07 |
| TheJulia | #topic Review subteam status reports | 15:07 |
| TheJulia | #link https://etherpad.opendev.org/p/IronicWhiteBoard | 15:07 |
| ajya | o/ | 15:07 |
| opendevreview | Dmitry Tantsur proposed openstack/metalsmith master: Add a CentOS job with legacy local boot https://review.opendev.org/c/openstack/metalsmith/+/841911 | 15:08 |
| dtantsur | this ^^^ is what we discussed re netboot removal | 15:08 |
| TheJulia | ack | 15:11 |
| TheJulia | It looks like just about everything has been updated. | 15:11 |
| TheJulia | Does anyone have any questions, comments in regarding to updates? | 15:12 |
| dtantsur | is anyone actually planning to work on active steps? | 15:12 |
| TheJulia | I'd like to, I just don't think I'm going to have time right now, I'm spinning a lot of plates and bugfixes at the moment | 15:13 |
| iurygregory | dtantsur, well we will probably remove I think (it was based on what we had in the etherpad from the zed PTG) | 15:13 |
| dtantsur | We probably shouldn't treat it as a priority if nobody is going to work on it :) | 15:13 |
| iurygregory | =) | 15:13 |
| iurygregory | yeah | 15:13 |
| TheJulia | I *might* be able to work on it, but my concern is even just reviews at this point | 15:16 |
| dtantsur | I may (fingers crossed) be able to return upstream, at least as far as reviews are concerned now | 15:16 |
| TheJulia | I likely wouldn't be abel to get the reviews until sometime int he fall | 15:16 |
| TheJulia | Well, that is good news | 15:17 |
| TheJulia | I guess we need to recognize this as a problem and work on a solution. Review Jamming seemed to really help when we did it | 15:17 |
| TheJulia | Anyway, that is outside the scope of updates | 15:17 |
| TheJulia | Onward? | 15:17 |
| iurygregory | yes | 15:18 |
| opendevreview | Dmitry Tantsur proposed openstack/metalsmith master: Add a CentOS job with legacy local boot https://review.opendev.org/c/openstack/metalsmith/+/841911 | 15:18 |
| TheJulia | #topic Deciding on priorities for the coming week | 15:19 |
| TheJulia | #link https://review.opendev.org/q/status:open+hashtag:ironic-week-prio | 15:19 |
| TheJulia | Does anyone have anything to add to this list? | 15:19 |
| iurygregory | https://review.opendev.org/c/openstack/ironic-python-agent/+/837039 maybe we can add the multipath patch | 15:20 |
| TheJulia | ++ | 15:20 |
| dtantsur | and the metalsmith patch above | 15:21 |
| TheJulia | Done | 15:22 |
| TheJulia | Anythine else? | 15:22 |
| TheJulia | /thine/thing/ | 15:22 |
| * dtantsur has too many backports that fails the CI, needs to burn through this list | 15:22 | |
| TheJulia | ditto | 15:23 |
| TheJulia | or conflicting backports I can't seem to get sorted | 15:23 |
| TheJulia | And small patches which bring value I should just finish | 15:23 |
| TheJulia | Odds of me sleeping on my LAX->AMS->BER trip?! | 15:23 |
| dtantsur | LAX->AMS possibly | 15:24 |
| TheJulia | heh | 15:24 |
| TheJulia | maybe.. | 15:24 |
| dtantsur | AMX->BER is what, 1.5 hours? :) | 15:24 |
| TheJulia | about | 15:24 |
| TheJulia | Anyway, onward! | 15:24 |
| TheJulia | arne_wiebalck: Do we have anything for the SIG this week? | 15:24 |
| arne_wiebalck | TheJulia: the video from last week is up already | 15:25 |
| arne_wiebalck | kudos to Steve once more | 15:25 |
| TheJulia | #topic Baremetal SIG | 15:25 |
| arne_wiebalck | apart from this NTR | 15:25 |
| TheJulia | Thanks stevebaker[m]! | 15:25 |
| TheJulia | Do we have our plans sorted for the session in Berlin? | 15:25 |
| arne_wiebalck | the *plan* is to have a look-back and have a meet & greet session | 15:26 |
| arne_wiebalck | but that is about all that happened until now :-D | 15:26 |
| TheJulia | arne_wiebalck: do we have an etherpad setup for notes? | 15:26 |
| TheJulia | ;) | 15:26 |
| TheJulia | okay | 15:26 |
| arne_wiebalck | not yet, I can do that ofc | 15:26 |
| TheJulia | I think that would be good, even just slightly open ended and we could include the "how to get in touch" details as well | 15:27 |
| arne_wiebalck | if there are suggestions what else to do, let us know | 15:27 |
| TheJulia | Okay! | 15:28 |
| TheJulia | Since we have nothing under RFE review, I guess we shall proceed to Open Discussion! | 15:28 |
| TheJulia | #topic Open Discussion | 15:28 |
| TheJulia | Someone has raised a Raid Cleaning Tempest test item? | 15:28 |
| JasonF | Hey, I wanted to let folks know -- a week from today I'll be starting a new position with G-Research out of the UK, and will have work-time to contribute to OpenStack again. | 15:29 |
| TheJulia | #link https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/841601/ | 15:29 |
| *** JasonF is now known as JayF | 15:29 | |
| TheJulia | Any volunteers for that topic? | 15:29 |
| TheJulia | JayF: \o/ woohooooooo | 15:29 |
| JayF | That's my contribution for open discussion :) I'll obviously make a point to start attending these again | 15:29 |
| kamlesh6808c | The Julia : Hi | 15:29 |
| kamlesh6808c | Currently, I am working on raid cleaning tempest test case implementation. This implementation is intended to work on physical baremetal servers. | 15:29 |
| kamlesh6808c | In order to have full coverage to test raid cleaning, as python drac client or any third party tool is not expected to use, we are planning to consume it via configuration file of tempest as multiline string. | 15:30 |
| kamlesh6808c | (eg. https://paste.openstack.org/show/bmguH5bdY1BO8x1NLKD1/).Please find linked WIP patch [https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/841601/]. | 15:30 |
| TheJulia | cool! | 15:30 |
| kamlesh6808c | My query is - Can we use multiline string in config, validate and utilize inside raid_cleaning implementation test Case in ironic tempest plugin as consumed in mentioned WIP patch. | 15:30 |
| ayoung | o/ | 15:30 |
| kamlesh6808c | Please make a note that schema which has been added in patch is still in work in progress. Schema would be furnished considering it should be reusable by all platform means. | 15:30 |
| kamlesh6808c | This has been tested and working fine as expected | 15:31 |
| dtantsur | as to multiline values, you could support escape sequences (\n) | 15:31 |
| TheJulia | kamlesh6808c: I suspect for multiline you ought to just point it at a file instead of use multiline | 15:31 |
| dtantsur | but if it's something long, linking to a file name instead my be better | 15:31 |
| TheJulia | multiline is *not* fun to debug | 15:31 |
| dtantsur | right | 15:31 |
| TheJulia | And even harder to use | 15:32 |
| TheJulia | if your an operator | 15:32 |
| TheJulia | (one of those, we know the code, but they often don't things) | 15:32 |
| rpioso | kamlesh6808c: Are escape characters needed in the configuration file? | 15:32 |
| kamlesh6808c | rpioso : not needed | 15:32 |
| TheJulia | A file would also just be simpler to code/test | 15:33 |
| TheJulia | FWIW | 15:33 |
| rpioso | dtantsur, TheJulia: The newly proposed storage inventory setting is JSON. | 15:33 |
| TheJulia | a file can be read in and used as json | 15:33 |
| TheJulia | multiline/dict entries in ini files are really just a major headache unless you absolutely know what your doing | 15:34 |
| TheJulia | And then require very verbose examples | 15:34 |
| TheJulia | files with contents loaded as json just seems.... way simpler to me. | 15:34 |
| * dtantsur agrees | 15:34 | |
| TheJulia | At least, that is my feeling | 15:34 |
| TheJulia | dtantsur: wait, we agree?!? | 15:35 |
| TheJulia | I'll alert the media :) | 15:35 |
| dtantsur | \o/ | 15:35 |
| rpioso | dtantsur, TheJulia: Thank you! | 15:35 |
| dtantsur | BRKNG | 15:35 |
| kamlesh6808c | ok | 15:35 |
| dtantsur | okay, I have a potentially nostalgic topic | 15:35 |
| dtantsur | what do you think about reviving ironic-webclient? | 15:35 |
| TheJulia | no objection | 15:35 |
| kamlesh6808c | thanks | 15:35 |
| TheJulia | as long as someone is going to do something with it :) | 15:36 |
| TheJulia | the fact we now support basic auth too... | 15:36 |
| dtantsur | I'm literally talking to a person who is interested in a lightweight UI for Ironic | 15:36 |
| rpioso | dtantsur, TheJulia: Other than that, is the proposed direction agreeable, at least at this moment :)? | 15:36 |
| dtantsur | potentially even in developing one | 15:36 |
| iurygregory | wow | 15:36 |
| iurygregory | nice! | 15:36 |
| dtantsur | in reality, there are a few Ironic projects that don't receive a lot of changes | 15:37 |
| dtantsur | ironic-ui, n-g-s, n-bm, metalsmith, even the clients | 15:37 |
| iurygregory | IPE cries | 15:37 |
| dtantsur | IPA too | 15:37 |
| dtantsur | IPE | 15:37 |
| dtantsur | meh | 15:37 |
| iurygregory | XD | 15:37 |
| dtantsur | anyway, if we can check a few times per cycle that it still works (and it should - our API is backward compatible), it should be fine for us to keep it in the list of deliverables | 15:38 |
| TheJulia | rpioso: yeah, I *think* | 15:38 |
| dtantsur | and then when I have my regular mess-with-other-stuff time, I can integrate it in Bifrost :) | 15:38 |
| TheJulia | ironic-ui has some dynamic field logic too which makes it not-necessary for us to do too much | 15:38 |
| TheJulia | Thank whats his name who was at Cray | 15:38 |
| dtantsur | Michael Krotscheck? | 15:39 |
| TheJulia | no | 15:39 |
| TheJulia | he was HPE and went to VMware | 15:39 |
| dtantsur | right | 15:39 |
| TheJulia | This guy literally did his wind down time writing ironic-ui | 15:39 |
| ayoung | Don't do it | 15:39 |
| TheJulia | before retiring | 15:39 |
| ayoung | Web Clients suck in time and space and energy | 15:39 |
| dtantsur | well.. I also think that everyone should use CLI | 15:39 |
| TheJulia | ayoung: I feel a battle royale sort of conflict on web uis | 15:40 |
| dtantsur | but we miss out on outreach opportunities by the virtue of not having it | 15:40 |
| ayoung | Heh...no my suggestion would be to start by just making ironic honor the HTML accepts header first and see what that getsy ou | 15:40 |
| ayoung | it is minimal effort and gets you an ugly (non complete with Horzion) ui | 15:40 |
| dtantsur | -2 let's not mix the concerns | 15:40 |
| TheJulia | yeah, different uses | 15:40 |
| dtantsur | a garbage auto-generated UI won't serve the needs I'm talking about | 15:40 |
| JayF | If we have someone who wants a web UI, and they're going to work on it, might as well have them work on ironic-webclient than $new_unofficial_thing | 15:40 |
| dtantsur | another benefit of ironic-webclient is that it exists | 15:41 |
| ayoung | dtantsur, I am dead serious here.. you are going to be building a UI that needs people with a different skill set. | 15:41 |
| TheJulia | well, it really becomes up to what they want to do and if it adds value to the overall community | 15:41 |
| JayF | I'd suggest maybe encouraging that person to do some dev docs/automation on how someone less-webby could maybe maintain it if folks disappear | 15:41 |
| dtantsur | ayoung: I am not. We have it. It's just deprecated and abandoned. | 15:41 |
| TheJulia | We don't have sufficent information, just initial ideas so far | 15:41 |
| TheJulia | and well, yeah, and the existing javascripty magic ui | 15:42 |
| dtantsur | Here is a thing, folks. Nobody is going to contribute to a project that is abandoned and removed from the repository. | 15:42 |
| TheJulia | which requires checking out HEAD~1 | 15:42 |
| TheJulia | ++ | 15:42 |
| dtantsur | If we think it's valuable, we need to keep the door open for potential adopters and contributors. | 15:42 |
| ayoung | whereas, HTML generation from the existig code actually helps on the REST side, not just as a WebUI. It also means that you don't have to explicitly enable every feature in the Javascript once you get it in the web API | 15:42 |
| iurygregory | https://opendev.org/openstack/ironic-webclient | 15:42 |
| ayoung | dtantsur, I am sure that you can make a really useful, valuable tool. | 15:43 |
| dtantsur | ayoung: to be clear: I'm talking about `cd ironic-webclient && git revert HEAD && git review` | 15:43 |
| TheJulia | lets disconnect these these discussions just a little. There is merit to both, but we should likely do this as a high bandwidth discussion | 15:43 |
| dtantsur | ayoung: you mean, the web does not exist? :) | 15:43 |
| ayoung | HEH | 15:43 |
| dtantsur | because virtually all web sites are not created the way you describe | 15:43 |
| ayoung | so, I would agree that a web UI would be usedful | 15:43 |
| TheJulia | guys | 15:43 |
| dtantsur | (I need to trust you that there is a single one that does) | 15:43 |
| ayoung | dtantsur, I know. I have alwasy been a heretic | 15:43 |
| * TheJulia notes ayoung likely has heretic shirts | 15:43 | |
| TheJulia | :) | 15:43 |
| ayoung | I also have experience on both sides of the wire, including years of writing sites when we could not trust Javascript | 15:44 |
| dtantsur | do we need a vote, iurygregory to make a call, or ...? | 15:44 |
| TheJulia | I'm not sure there is a single definitive path from this discussion | 15:44 |
| TheJulia | there are merits, there are positives, there are negatives | 15:44 |
| * dtantsur disagrees | 15:44 | |
| iurygregory | well, I'm totally ok if people want to bring back the webclient | 15:44 |
| ayoung | dtantsur, if you feel strong enough about it, and want to put in the effort to maintain it, I say go for it. My advice was directed at you personally :) | 15:45 |
| iurygregory | I do see more positives than negatives | 15:45 |
| dtantsur | ayoung: well.. I don't want to write a new UI for Ironic, that's for sure. | 15:45 |
| ayoung | i.e do you want to put React on your resume? | 15:45 |
| dtantsur | no matter which method, not with React, nor with HTML | 15:45 |
| dtantsur | I also don't believe operators should use GUI | 15:46 |
| ayoung | not with a fox not in a box | 15:46 |
| dtantsur | but I think we're missing some outreach opportunities by not having a quick-start GUI | 15:46 |
| TheJulia | ++ | 15:46 |
| dtantsur | and we have two of them, both are web based | 15:46 |
| dtantsur | if we have one written in Qt and C++, I'd take it as well | 15:46 |
| dtantsur | even though I hate C++ | 15:46 |
| TheJulia | That is one of the biggest frustration points I've traditionally heard. A lack of a UI to "cut one's teeth on" | 15:46 |
| ayoung | so the one thing I would recommend thinking through is how to map from the UI to the CLI and web API | 15:46 |
| ayoung | how do you take someone from newbie to power user | 15:46 |
| TheJulia | you know, steve did a thing as a mobile app one day for fun | 15:47 |
| ayoung | Speaking of heretics | 15:47 |
| dtantsur | for me, familiarity with concepts is an important thing | 15:47 |
| dtantsur | I've literally recently had a serious customer escalation because they did not understand our concepts | 15:47 |
| dtantsur | (our = metal3 in that case) | 15:47 |
| TheJulia | ouch | 15:48 |
| dtantsur | once you understand the domain (nodes, states, ports, other stuff), mapping buttons to magical spells is a no brainer | 15:48 |
| ayoung | So, as I sit here working through a baremetal install on an alternate platform and learning concpets from first principals, I can agree with you dtantsur that a web UI would help | 15:48 |
| * TheJulia remembers aeva walking into a room with something the last time we had a deep UI discussion in ?vancouver? | 15:49 | |
| dtantsur | eons ago | 15:49 |
| ayoung | I think that was 3 job titles ago for me | 15:49 |
| TheJulia | this is one of those topics :) | 15:50 |
| ayoung | So...one more issue? | 15:50 |
| ayoung | Is anyone working on aarch or other architecture support besides me? | 15:50 |
| ayoung | aarch64 | 15:50 |
| dtantsur | I'll take an action items to look at ironic-webclient to see 1) how usable it is, 2) how terrible the code is to maintain | 15:50 |
| dtantsur | ayoung: there were people in RH OSP that made TripleO deploy on ARM | 15:51 |
| TheJulia | ayoung: someone at RH is supporting it for her work | 15:51 |
| TheJulia | cbouchar: o/ | 15:51 |
| TheJulia | ayoung: you and cbouchar should like... chat :) | 15:51 |
| dtantsur | ayoung: we had a cycle priority in Yoga to create a simple ARM job with Bifrost - did not happen. You could be The Chosen One ;) | 15:51 |
| JayF | If we do re | 15:51 |
| JayF | whoops | 15:52 |
| ayoung | TheJulia, the thing I am realizing looking at the code is that it would be a better refactoring to support multiple archs for a single deployment than to deploy assuming the controller and nodes are all the same arch | 15:52 |
| ayoung | do re me fa? | 15:52 |
| TheJulia | ayoung: bifrost specifically | 15:52 |
| TheJulia | ? | 15:52 |
| dtantsur | ayoung: ++ a lot | 15:52 |
| TheJulia | ayoung: I agree w/r/t bifrost a lot | 15:52 |
| ayoung | I thought JayF was channeling his inner Maria | 15:52 |
| TheJulia | ayoung: he might?! | 15:52 |
| * TheJulia does not know | 15:53 | |
| dtantsur | multi-arch is a thing for sure | 15:53 |
| * TheJulia is glad this is a super lively meeting | 15:53 | |
| ayoung | aso...are there old plans for multi-arch I can look at? | 15:53 |
| JayF | ayoung: I just have this problem I can't solve | 15:53 |
| ayoung | JayF, a willothewhisp? | 15:53 |
| dtantsur | ayoung: there are bits in the code | 15:53 |
| TheJulia | for bifrost, I don't really beleive so, more a "does this seem to work and is it easy". At least, beyond freagments of code/changes | 15:54 |
| dtantsur | ayoung: https://github.com/openshift/ironic-image/commit/fbf34c9757bf8bc9dc602823fed28c43f0b3dbb4 may be handy | 15:54 |
| JayF | ayoung: Maria! | 15:54 |
| ayoung | dtantsur, thanks you | 15:54 |
| dtantsur | ayoung: we'll need to start publishing ARM images for IPA | 15:56 |
| ayoung | is jeffdyoung in irc? | 15:56 |
| TheJulia | So, we have 4 minutes left | 15:56 |
| dtantsur | e kinda do actually: https://tarballs.opendev.org/openstack/ironic-python-agent-builder/dib/ | 15:56 |
| TheJulia | #topic Who is running the next meeting? | 15:56 |
| dtantsur | but these are experimental | 15:56 |
| iurygregory | I can run o/ | 15:56 |
| TheJulia | Awesome! | 15:57 |
| TheJulia | Our fearless leader will run our next meeting! | 15:57 |
| TheJulia | Thanks everyone! | 15:57 |
| TheJulia | #endmeeting | 15:58 |
| opendevmeet | Meeting ended Mon May 16 15:58:09 2022 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:58 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2022/ironic.2022-05-16-15.01.html | 15:58 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2022/ironic.2022-05-16-15.01.txt | 15:58 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2022/ironic.2022-05-16-15.01.log.html | 15:58 |
| * iurygregory goes for lunch brb | 15:58 | |
| * TheJulia signs... time for another meeting in a moment | 15:58 | |
| TheJulia | sighs | 15:58 |
| sdanni | Hi all! I have a question w.r.t ironic HA. HA ironic assigns each node to a specific conductor, and the ipxe files in pxelinux.cfg generated for each node during introspection or cleaning only appear on that specific controller. But when we use neutron dhcp, multiple neutron dhcp agents on different controllers can respond to a dhcp request, and we're running into cases where that results in the node being pointed at | 15:59 |
| sdanni | a conductor that does not contain those ipxe files. Then we get "pxelinux.cfg/<node_nic_mac_addr> no such file or directory". Any advice on fixing this issue? | 15:59 |
| dtantsur | sdanni: hmmm, shouldn't be happening. DHCP should point at the Ironic's TFTP server, not at Neutron.. | 16:00 |
| TheJulia | In *other* news, I've purchased a garmet suitcase | 16:01 |
| TheJulia | I feel like an adult finally | 16:01 |
| dtantsur | wow, and how does it feel? | 16:02 |
| rpioso | My grad school age children have shared with my wife and me that adulting is hard. | 16:04 |
| JayF | Oh wow, this seems like a really bad idea https://review.opendev.org/c/openstack/ironic-python-agent/+/841220 -- like I know it's a projectwide governance thing | 16:04 |
| JayF | but we need to consider lowering the distance between eventlet versions we support | 16:04 |
| JayF | I've literally lost a month of my life to things broken in different ways at different eventlet versions in our supported range | 16:04 |
| JayF | (and that change woulda merged if it wasn't for the LC job) | 16:05 |
| TheJulia | it feels bizzar... | 16:07 |
| JayF | If there's no objection from the rest of the group, I was thinking about looking at making python 3.10 support in ironic and ironic-dependant-tools first-class | 16:08 |
| JayF | as a potential first project for getting back involved | 16:08 |
| JayF | e.g. I know py 3.10 jobs in IPA are non-voting, and that DIB doesn't support 3.10 at all right now, and I'd like to fix that | 16:09 |
| sdanni | dtantsur: yeah, but there are three tftp servers, one on each node. They can all respond to a node. | 16:11 |
| dtantsur | sdanni: tftp servers don't respond in the same sense as DHCP servers | 16:12 |
| dtantsur | Neutron passes the IP address of the TFTP server to use as part of its response | 16:12 |
| dtantsur | JayF: I don't think you'll face any objections when working on supporting a new Python version :) | 16:12 |
| dtantsur | as far as I am concerned, 3.11 is fine too :) | 16:13 |
| JayF | I obviously have to talk to my new boss when I start next week, but I hope to have the opportunity to work on sorta... boring pieces like this? | 16:13 |
| JayF | I won't be running Ironic so I feel a little weird working directly on features or design, but I know there's always CI/python version type housekeeping to do | 16:13 |
| opendevreview | Dmitry Tantsur proposed openstack/metalsmith master: Add a CentOS job with legacy local boot https://review.opendev.org/c/openstack/metalsmith/+/841911 | 16:17 |
| dtantsur | JayF: true | 16:17 |
| sdanni | dtantsur: does it mean that only the controller assigned to the baremetal node will generate ipxe files and its tftp will respond to the dhcp requests? | 16:18 |
| dtantsur | sdanni: that's the plan. I'm not sure why it does not work in your case. | 16:18 |
| dtantsur | you may sniff the DHCP traffic with tcpdump to see which options are sent | 16:18 |
| sdanni | sounds good! | 16:19 |
| tzumainn | dtantsur, hi! quick clarification: what service is supposed to be responding the dhcp requests in the use case sdanni is talking about? we've only seen neutron dhcp respond | 16:25 |
| dtantsur | tzumainn: neutron, yes | 16:38 |
| dtantsur | but it does not respond to tftp | 16:38 |
| tzumainn | dtantsur, okay, yep, that makes sense - and neutron guides the node towards the correct http server with the node's mac address file under /var/lib/ironic/httpboot? | 16:42 |
| dtantsur | or the correct tftp server - depending on what is available on the machine | 16:43 |
| gmarcy_ | hello. new to the chat but long time ironic user | 16:46 |
| dtantsur | hi again gmarcy_ :) | 16:47 |
| tzumainn | dtantsur, okay, so maybe the part I'm not understanding is how neutron works here - how does neutron guide the node to the correct conductor? | 16:48 |
| dtantsur | tzumainn: it sends a DHCP option that looks like this: https://github.com/metal3-io/ironic-image/blob/main/ironic-config/dnsmasq.conf.j2#L31 | 16:49 |
| dtantsur | if iPXE is already running, it can provide a URL instead: https://github.com/metal3-io/ironic-image/blob/main/ironic-config/dnsmasq.conf.j2#L21 | 16:49 |
| tzumainn | dtantsur, okay - and the Ironic IPs or Ironic URL host has to be specific to a particular ironic conductor? | 16:51 |
| dtantsur | tzumainn: yes. in case of neutron, ironic provides DHCP options to it. | 16:51 |
| dtantsur | (it's a bit different for standalone as you see) | 16:51 |
| tzumainn | dtantsur, ah, okay! so in the HA case: Ironic receives a request to clean a node, it generates various files on the node's matching conductor, then tells neutron to configure dhcp to guide the node to that conductor - does that sound right? | 16:54 |
| dtantsur | yep | 16:54 |
| tzumainn | okay, thanks! | 16:54 |
| TheJulia | Question of the day: how many meetings will I have | 16:59 |
| dtantsur | all of them? | 17:01 |
| dtantsur | on this positive note I'm wishing you a nice evening o/ | 17:04 |
| cbouchar | ayoung: Yes I'm working on multiarch but I've been temporarily derailed doing work on an the older project. | 17:20 |
| cbouchar | ayoung: I'm mostly a user at the moment. | 17:20 |
| * TheJulia opens twitter, finds inappropriate private message some rando sent on there, and just closes it again. | 17:34 | |
| TheJulia | Now... where did I put my brain | 17:35 |
| ayoung | cbouchar, do you have design ideas how to proceed? | 17:37 |
| ayoung | Is it all bifrost, or some Ironic | 17:37 |
| ayoung | So...who issues the certificates, and why is it issuing a certificate that is not yet valid? | 17:41 |
| ayoung | certificate verify failed: certificate is not yet valid (_ssl.c:1131)'')))' | 17:41 |
| ayoung | I did a quick check via ipmi and the clocks seem like they areclose (within a second) of each other | 17:42 |
| TheJulia | not yet valid *really* makes me think clocks are out of sync | 18:02 |
| TheJulia | well... ipmi is the BMC's clock | 18:02 |
| TheJulia | not OS's clock | 18:03 |
| ayoung | So...how would one sync up the clocks in a cluster? Say I ran an ntpd on the controller, wouldthe cleaningimage then be responsible as the ntp client for doingf the hardware clock adjustment? And does that mean I need a customer IPA image? | 18:37 |
| ayoung | cbouchar, do you have an aarch64 IPA build approach I could borrow? | 18:39 |
| TheJulia | ayoung: cbouchar is kickstarting machines | 18:39 |
| TheJulia | ayoung: ipa-ntp-server pxe append parameters option | 18:40 |
| TheJulia | As long as chrony or ntpdate is present it should force sync the clock | 18:41 |
| ayoung | meaning I can add that option to somewhere in the PXE boot process.... | 18:42 |
| TheJulia | Yes | 18:47 |
| ayoung | Um...where do I put that? | 18:57 |
| * ayoung never studied | 18:57 | |
| ayoung | https://getyarn.io/yarn-clip/71a7a734-ea7f-42ec-9ed1-bd6ef9c0a2c3 | 18:57 |
| ayoung | pxelinux.cfg/default ? | 19:00 |
| ayoung | https://docs.openstack.org/ironic/pike/install/include/kernel-boot-parameters.html | 19:01 |
| TheJulia | [pxe]pxe_append_params | 19:10 |
| opendevreview | Merged openstack/ironic-python-agent master: Drop lower-constraints.txt and its testing https://review.opendev.org/c/openstack/ironic-python-agent/+/841220 | 19:34 |
| opendevreview | Steve Baker proposed openstack/bifrost master: Support PXE network boot with grub https://review.opendev.org/c/openstack/bifrost/+/807220 | 21:15 |
| opendevreview | Steve Baker proposed openstack/bifrost master: Switch grub network boot from TFTP to HTTP https://review.opendev.org/c/openstack/bifrost/+/842008 | 22:34 |
| TheJulia | ++ | 23:10 |
| TheJulia | we really need to just do the httpboot stuffs as well given the newer UEFI images support it | 23:11 |
| TheJulia | ... and we basically have the code... we'e just not used the correct magical incantation | 23:11 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!