| opendevreview | Harald Jensås proposed openstack/metalsmith master: list_instances - cache allocations https://review.opendev.org/c/openstack/metalsmith/+/873048 | 00:42 |
|---|---|---|
| hjensas | TheJulia: stevebaker[m]: I got it from 6min+ to ~55seconds - metalsmith list (250 nodes) with ^^ patches. | 00:44 |
| stevebaker[m] | Hey thats a good improvement | 00:52 |
| TheJulia | Nice! | 01:09 |
| *** rcastillo|rover is now known as rcastillo | 03:59 | |
| *** JasonF is now known as JayF | 04:12 | |
| rpittau | good morning ironic! o/ | 08:18 |
| janders | hey rpittau and Ironic o/ | 08:39 |
| rpittau | hey janders :) | 08:39 |
| opendevreview | Harald Jensås proposed openstack/metalsmith master: list_instances - cache allocations https://review.opendev.org/c/openstack/metalsmith/+/873048 | 09:08 |
| opendevreview | Mark Goddard proposed openstack/networking-generic-switch master: WIP: Attempt to batch up cmds https://review.opendev.org/c/openstack/networking-generic-switch/+/743283 | 09:51 |
| opendevreview | Mark Goddard proposed openstack/networking-generic-switch master: Support batching up commands https://review.opendev.org/c/openstack/networking-generic-switch/+/743283 | 10:30 |
| opendevreview | Mark Goddard proposed openstack/networking-generic-switch master: Add Dummy device type https://review.opendev.org/c/openstack/networking-generic-switch/+/873098 | 10:38 |
| dtantsur | kubajj: API methods do not normally raise, they return an HTTP error. Look above for examples, you need expect_errors=True. | 11:25 |
| opendevreview | Riccardo Pittau proposed openstack/metalsmith master: [WIP] Add centos9 based job https://review.opendev.org/c/openstack/metalsmith/+/869374 | 11:31 |
| opendevreview | Jakub Jelinek proposed openstack/ironic master: Erase swift inventory entry on node deletion https://review.opendev.org/c/openstack/ironic/+/871394 | 12:13 |
| kubajj | TheJulia, dtantsur: Oh, I see. Thanks for help | 12:14 |
| opendevreview | Merged openstack/bifrost master: Fix jinja ansible lint error https://review.opendev.org/c/openstack/bifrost/+/866137 | 12:14 |
| kubajj | JayF: is it this one? https://review.opendev.org/c/openstack/ironic/+/866235 | 13:05 |
| iurygregory | good morning Ironic | 13:52 |
| *** tosky_ is now known as tosky | 14:19 | |
| scottsol | Is the baremetal SIG going ahead today? | 14:36 |
| JayF | scottsol: aye | 14:38 |
| JayF | scottsol: also, I didn't know you wre in IRC land, assuming you are who I think you are :) | 14:39 |
| scottsol | do you even OpenStack if you're not in IRC land ;) | 14:39 |
| JayF | iurygregory: TheJulia: dtantsur: I'm in BM SIG zoom if we wanna like, audio check and such | 14:39 |
| JayF | scottsol: surprisingly, yes | 14:40 |
| opendevreview | Will Szumski proposed openstack/bifrost stable/xena: Create /run/ironic runtime directory when service starts https://review.opendev.org/c/openstack/bifrost/+/873117 | 14:42 |
| JayF | kubajj: that chain, the first one is a step behind it | 14:43 |
| TheJulia | I'm in a required conference session atm :( | 14:44 |
| JayF | "required" is just a check against your will saving throw ;) | 14:44 |
| dtantsur | JayF: I'll join in a few | 14:46 |
| TheJulia | heh | 14:47 |
| opendevreview | Will Szumski proposed openstack/bifrost stable/xena: Create /run/ironic runtime directory when service starts https://review.opendev.org/c/openstack/bifrost/+/873117 | 14:47 |
| JayF | Baremetal SIG starting in 3 minutes; https://ironicbaremetal.org/blog/baremetal-sig-2023q1/ can get you to the zoom link | 14:58 |
| opendevreview | Oliver Walsh proposed openstack/sushy-tools master: Fix libvirt forbidden error in set_boot_mode https://review.opendev.org/c/openstack/sushy-tools/+/873123 | 15:28 |
| opendevreview | Oliver Walsh proposed openstack/sushy-tools master: Fix libvirt forbidden error in set_boot_mode https://review.opendev.org/c/openstack/sushy-tools/+/873123 | 15:41 |
| opendevreview | Oliver Walsh proposed openstack/sushy-tools master: Fix libvirt forbidden error in set_boot_mode https://review.opendev.org/c/openstack/sushy-tools/+/873123 | 15:42 |
| iurygregory | no talk for OIS 2023 .-. | 15:59 |
| iurygregory | maybe i should have submitted to another track (instead of Hybrid/Private Cloud) | 16:11 |
| JayF | I think also there was a push to get more new talks in? | 16:15 |
| JayF | So it's the case where being a common person to have done talks in the past set you up to be more likely to fail | 16:15 |
| JayF | but IDK; I wasn't a chair for hybrid/priv cloud | 16:15 |
| iurygregory | well, I only had two talks in OIS (Virtual in 2021/ Berlin 2022) XD | 16:17 |
| iurygregory | let's wait for the Forum sessions to see (but I probably won't be able to travel to Vancouver with only Forum sessions .-.) | 16:18 |
| TheJulia | Is there a sig zoom link still? | 16:24 |
| rpittau | TheJulia: it's in https://lists.openstack.org/pipermail/openstack-discuss/2023-February/032031.html | 16:24 |
| NobodyCam | Good morning Ironic Folks, | 16:56 |
| NobodyCam | and OFC | 16:56 |
| NobodyCam | happy hump day | 16:56 |
| NobodyCam | ;) | 16:56 |
| rpittau | hey NobodyCam :) | 17:10 |
| opendevreview | Oliver Walsh proposed openstack/sushy-tools master: Fix libvirt forbidden error in set_boot_mode https://review.opendev.org/c/openstack/sushy-tools/+/873123 | 17:31 |
| rpittau | good night! o/ | 17:32 |
| JayF | dtantsur: where does this recording go? | 17:32 |
| JayF | dtantsur: I assume there's some official openstacky place it can be uploaded? | 17:33 |
| dtantsur | arne_wiebalck: ^^^ | 17:33 |
| samuelkunkel[m] | s3? (sorry, could not resist) :D | 17:33 |
| JayF | after it converts, I'll upload to gdrive and get a share link to whoever? | 17:33 |
| JayF | samuelkunkel[m]: you better watch out what you ask for; the product I started into Ironic with was predicated on how predatory the AWS pricing model is | 17:33 |
| JayF | samuelkunkel[m]: so I'm about one or two more snarks away from giving you my dissertation on cloud costs that scale non-linearly in a designed, predatory way ;) | 17:34 |
| JayF | lol | 17:34 |
| samuelkunkel[m] | uff | 17:34 |
| samuelkunkel[m] | That sounds frightening | 17:34 |
| JayF | AWS, cheap for startups, then it hockey sticks | 17:34 |
| JayF | unlike Bare Metal Servers(tm) | 17:34 |
| JayF | etc etc | 17:34 |
| samuelkunkel[m] | we can have a chat about that in vancouver :) | 17:35 |
| JayF | clearly that was a great sales pitch delivered effectively since nobody from Rackspace still works on Ironic LOL | 17:35 |
| samuelkunkel[m] | lol | 17:35 |
| JayF | arne_wiebalck: dtantsur: I shared the drive folder with the video to both of you; if there's anything further I should do please let me know | 17:49 |
| jrosser | when updating the controllers, like at major openstack version upgrade, is it the right thing to just bulk update the deploy kernel/initrd uuid on all the nodes to point to the new ones for the new release running on the controller? | 17:54 |
| jrosser | regardless of the deploy state of the nodes | 17:54 |
| JayF | Honestly, when working at scale, I tried really, really hard to avoid setting the deploy kernel/initrd per node unless I had to | 17:59 |
| JayF | the answer to your question is "yes, they should be updated" | 17:59 |
| jrosser | but it’s version specific? | 18:00 |
| jrosser | well, described as, anyway | 18:01 |
| JayF | the initrd in this case includes the actual code for IPA | 18:01 |
| JayF | so the python code inside needs an upgrade just like the service code does | 18:01 |
| JayF | although generally unless a release note says specifically otherwise, there's some backwards compat | 18:01 |
| jrosser | right, so I did my upgrade and switched all the uuid over | 18:03 |
| jrosser | but what made me think that I might be doing something wrong was that didn’t make sense for $lots nodes | 18:04 |
| JayF | you can set deploy kernel/initrd in config | 18:04 |
| JayF | rather than just per node | 18:04 |
| jrosser | oh right ok | 18:04 |
| jrosser | even for multi arch? | 18:04 |
| JayF | I don't know how that feature interacts with multi-arch | 18:04 |
| JayF | if we've introduced a scenario where you need to do N updates for N nodes to update the deploy kernel/initrd, that's a UX pain point we'd probably be keen to remove | 18:05 |
| JayF | e.g. by making a cpu-arch-aware default or something | 18:05 |
| jrosser | I thought that some kind of template would be interesting | 18:06 |
| jrosser | where you update once in the template and reference that in the node | 18:06 |
| jrosser | but that’s just random musing :) | 18:06 |
| JayF | might be a good thing for PTG? IDK | 18:06 |
| JayF | but I could see setting deploy_kernel: `arm_default` on a node | 18:07 |
| JayF | and then being able to set deploy_kernel {'arm_default': UUID} in a config | 18:07 |
| jrosser | yes, that sort of thing | 18:10 |
| JayF | I think iurygregory did something similar for external_url recently? but IDR for sure | 18:14 |
| JayF | please do add something to the PTG page about this -> https://etherpad.openstack.org/p/ironic-bobcat-ptg | 18:15 |
| iurygregory | yup I did | 18:15 |
| TheJulia | Hmm interesting | 18:15 |
| iurygregory | https://review.opendev.org/c/openstack/ironic/+/852234/2 | 18:16 |
| TheJulia | It seems like we need a setting by arch default in the config | 18:17 |
| TheJulia | Like what we do with pxe loader | 18:17 |
| JayF | I don't think a more generic mechanism is a bad idea either | 18:17 |
| JayF | because I had cases where I needed to set explicitly different ramdisks by hardware vendor, as well | 18:18 |
| TheJulia | By arch can be that way as well because it is just tag matching, really | 18:18 |
| JayF | ack | 18:20 |
| jrosser | I think adding the debug user to the deploy image is another use case | 18:21 |
| jrosser | where you might want to opt a single node into that | 18:22 |
| JayF | that's one of the exact cases single-node overrides were made for | 18:22 |
| jrosser | but you also need to manage the versions of the debug image across upgrades so the same indirection would be great there | 18:23 |
| * JayF put that item in PTG planning | 18:31 | |
| kubajj | What is oslo.db? Is it just some kind of a wrapper around sqlalchemy? | 18:33 |
| JayF | anytime you see "oslo." | 18:35 |
| JayF | that means "openstack library" more or less | 18:36 |
| JayF | (not that you can't use oslo libs in other projects; but oslo is the name of the project that does cross-project libs) | 18:36 |
| JayF | I honestly don't know exactly what we get from oslo.db because I've not dug into it personally | 18:36 |
| JayF | but I think it's more than just "wrapper around sqla" but IMBW | 18:36 |
| kubajj | JayF: thanks, good to know | 18:37 |
| TheJulia | mostly just a wrapper with some helper utilities | 18:38 |
| TheJulia | "mostly" | 18:38 |
| jrosser | JayF: thanks for putting that on the ptg notes, I’m on a phone irc client just now so it’s tricky :) | 18:41 |
| JayF | Whoa, tripleo is gone | 18:47 |
| TheJulia | It is certainly a complex topic | 18:57 |
| Sandzwerg[m] | Jrosser: we've used IPA images that where some releases newer than our ironic version. Worked well for us | 19:05 |
| TheJulia | we've strived to not introduce breaking changes. The only one I can actually think of is the agent token stuff, which we explicitly communicated as we felt it was the right thing to do | 19:09 |
| JayF | and that one was opt-out if you built your own image iirc | 19:16 |
| TheJulia | ... I *think* for one major release | 19:17 |
| TheJulia | at which point, the handling behavior went towards "mandatory" | 19:17 |
| TheJulia | unfortunately, it feels like that was ages ago | 19:18 |
| arne_wiebalck | stevebaker[m]: I uploaded the SIG video from today ... whenever you have a moment to do your editing magic :) | 20:48 |
| stevebaker[m] | arne_wiebalck: OK I'll get on to that. I *really* want to record a musical intro for these videos | 20:54 |
| arne_wiebalck | stevebaker[m]: the sound of a dial-up modem :-D | 20:55 |
| ashinclouds[m] | I love that idea | 20:57 |
| stevebaker[m] | yeah I was thinking of incorporating a chiptunes vibe | 21:03 |
| TheJulia | I was thinking of how awesome the v.92 ping/ping sound was | 21:04 |
| TheJulia | Personally | 21:04 |
| JayF | if it's not metal | 21:05 |
| JayF | what the hell are you doing | 21:05 |
| JayF | BEAR METAL | 21:05 |
| TheJulia | Heh | 21:05 |
| * TheJulia heads towards the metal tubes | 21:06 | |
| JayF | have a good flight o/ | 21:06 |
| TheJulia | I will try… over 3 hours until takeoff | 21:06 |
| TheJulia | Sooo many hours until the sleeeeeps | 21:06 |
| stevebaker[m] | TheJulia: \o | 21:08 |
| * TheJulia raises eyebrow | 21:09 | |
| * stevebaker[m] listens to modem sounds on youtube | 21:09 | |
| TheJulia | Heh | 21:10 |
| TheJulia | “Next up on the Bare Metal Sig! The whistling to the 300 baud modem contest!” | 21:11 |
| * TheJulia is technically in a rolling metal tube at the moment of the variety which can change lanes | 21:11 | |
| JayF | this is totes going to help the openstack perception of being outta date | 21:17 |
| JayF | lol | 21:17 |
| TheJulia | Okay, 60hz and various harmonics of a 3 phase transformer? | 21:19 |
| TheJulia | Or 50hz | 21:20 |
| * TheJulia shows herself the door to the metal tube… soon :) | 21:20 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!