| shermanm | A thought, but is it worth revisiting kexec for this? Vmedia boot a kernel and initramfs with the right stuff to connect to nvme over tcp, send keys or whatever through a config drive, and then kexec into the volume? I guess you could skip it if the kernel and initramfs always match | 02:32 |
|---|---|---|
| cardoe | shermanm: I can tell you the answer on the performance on that one iDRAC 9 | 04:34 |
| *** kurtyoon is now known as Guest2830 | 06:23 | |
| *** dmellado4717259121 is now known as dmellado471725912 | 10:11 | |
| TheJulia | shermanm: I'm not sure you can populate IBFT and expect the kexec kernel to read it. | 12:55 |
| TheJulia | cardoe: on idrac9 meaning, the through the network device function path? | 12:56 |
| TheJulia | Also, good morning! | 12:57 |
| cardoe | TheJulia: through the circular buffer shermanm mentioned | 12:58 |
| cardoe | We seem to have some error that's breaking openstacksdk tests.. https://zuul.opendev.org/t/openstack/build/a7a41b3905b14aed8460fac307970134 | 12:59 |
| TheJulia | oh, for vmedia as well | 13:00 |
| TheJulia | clif: you around? | 13:12 |
| clif | TheJulia: omw | 13:15 |
| shermanm | TheJulia: with kexec, the thought is that we could take secrets, config, etc, stuff them into a cpio archive, and append them onto the end of the volume's initramfs when handing over | 13:44 |
| shermanm | (Caveat, this comes from very naive searches, I'm not confident) | 13:44 |
| shermanm | cardoe: that would be quite helpful actually, w.r.t. performance. Turns out I have a fleet of r630,r640,r650 and related systems that I can run tests on with different bios/bmc/kernel configs | 13:45 |
| cardoe | it's the same bit bang interface into that buffer | 14:05 |
| cardoe | the buffer only exists to not require you to be connected at all times otherwise you lose the data | 14:05 |
| shermanm | :( | 14:24 |
| shermanm | there's also non-blocking console writes in the kernel now with nbcon, but that only helps post-boot | 15:02 |
| JayF | nbcon is still a framebuffer-adjacent thing that needs graphics, yeah | 15:05 |
| JayF | FWIW first TC topic @ 1600 is GUIs in OpenStack | 15:32 |
| TheJulia | wheeeeeeeeeeee | 15:36 |
| cardoe | pending TBN patches: https://review.opendev.org/c/openstack/ironic/+/1010775 https://review.opendev.org/c/openstack/ironic/+/1009783 https://review.opendev.org/c/openstack/ironic/+/1009782 https://review.opendev.org/c/openstack/ironic/+/1009337 https://review.opendev.org/c/openstack/ironic/+/1009360 | 15:38 |
| cardoe | https://review.opendev.org/c/openstack/ironic/+/1009781 https://review.opendev.org/c/openstack/ironic/+/1009780 | 15:39 |
| cardoe | https://review.opendev.org/c/openstack/ironic/+/1008898 | 15:40 |
| JayF | toss a topic or hashtag on em, perhaps? | 15:41 |
| cardoe | They're in ironic-week-prio but I can put a topic of like tbn-fixes | 15:46 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: fix case where neutron ports are not tracked with the baremetal port https://review.opendev.org/c/openstack/ironic/+/985050 | 15:53 |
| cardoe | shermanm: ^ that's the rebased again fix for ironic not tracking ports | 15:53 |
| shermanm | tx! | 15:56 |
| shermanm | btw, did the spec for "pluggable http server for ironic", exist yet? | 15:56 |
| cardoe | Don't think so | 15:57 |
| shermanm | so, our crazy use case: We run a centralized ironic/nova/etc control plane somewhere. We're working with https://nrp.ai/. They hand us BMC credentials and a vpn for access to N baremetal server BMCs, around the country, and we enroll those nodes into ironic, use virtual media boot to provision them and hand off to end users to use the machines. | 16:00 |
| shermanm | so that a user can do something like "try experiment X on expensive GB300 node Y for 1 day", and then it gets wiped and rejoins its parent cluster | 16:01 |
| shermanm | generally trying to drastically lower the barrier to "hey you can use this expensive thing for a few days" | 16:01 |
| JayF | shermanm: dtantsur had a very intense RFE about it | 16:02 |
| JayF | shermanm: oh, we went a different direction than that RFE, nevermind | 16:02 |
| shermanm | but in my head the same design could let vendors temporarily contribute hardware for testing, driver compat, etc, even if they don't have the manpower / desire to actually join up to zuul | 16:02 |
| shermanm | (other people's hardware makes life difficult) | 16:04 |
| shermanm | w.r.t. the snapshot and in-band service steps code/spec: is there an existing way I should be getting credentials? since this will need to create a glance image owned by the user | 16:06 |
| cardoe | https://review.opendev.org/q/topic:%22tbn-fixes%22 are the TBN fixes | 16:09 |
| cardoe | adamcarthur5: https://review.opendev.org/c/openstack/ironic/+/1003204 if you split that patch from the series we can land that one. | 17:00 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: Mask secret step arguments in conductor logs https://review.opendev.org/c/openstack/ironic/+/1010980 | 17:03 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 17:28 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 17:31 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 17:34 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 17:41 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: Fix ambiguous owner when listing allocations by node https://review.opendev.org/c/openstack/ironic/+/1010992 | 18:00 |
| cardoe | TheJulia: JayF: ^ the openstacksdk functional test is actually finding a bug of ours | 18:01 |
| JayF | can we run them? | 18:01 |
| cardoe | You know Zuul better than me if that's a circular loop | 18:03 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 18:37 |
| TheJulia | https://pypi.org/project/ironic-interfaces <--- Queue eyebrow raise | 18:39 |
| JayF | mnaser: ^ is who published that | 18:41 |
| JayF | TheJulia: you are the person who I would've pinged if I saw that. So hey, Julia, it looks like someone is misusing our trademark publically. | 18:41 |
| JayF | from https://github.com/vexxhost/ironic-interfaces | 18:41 |
| TheJulia | I can raise that to the staff for trademark eval, I think it would fall under fair use plugining though | 18:43 |
| TheJulia | but the lack of details is more so the eyebrow raise | 18:44 |
| TheJulia | its definitely some driver stuff | 18:44 |
| JayF | mnaser: if we ask nicely can you publish that under a namespace, preferably not including *ironic* (but at least not *starting with* ironic*) | 18:44 |
| JayF | mnaser: I don't want folks to be confused | 18:44 |
| JayF | TheJulia: https://www.youtube.com/shorts/KHBl2kcGZjM (related to final chat message in PTG) | 18:46 |
| * TheJulia points herself at the door | 18:47 | |
| TheJulia | ;) | 18:47 |
| TheJulia | lol | 18:47 |
| cardoe | I was pretty ruthless in reviewing ironic-week-prio. There was a lot of stuff in there that's been bit rotted so I cleaned it out. I've reviewed everything except the ironic-networking pieces. | 20:17 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 21:11 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Fix Cisco NX-OS trunk port configuration https://review.opendev.org/c/openstack/networking-generic-switch/+/1010987 | 21:16 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Add Cisco NX-OS bond trunk support https://review.opendev.org/c/openstack/networking-generic-switch/+/1011015 | 21:54 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Add Cisco NX-OS bond trunk support https://review.opendev.org/c/openstack/networking-generic-switch/+/1011015 | 21:56 |
| opendevreview | Bartosz Bezak proposed openstack/networking-generic-switch master: Add Cisco NX-OS bond trunk support https://review.opendev.org/c/openstack/networking-generic-switch/+/1011015 | 21:57 |
| TheJulia | Hey folks, I've posted a summary to https://etherpad.opendev.org/p/ironic-2027.1-ptg-summary I'm planning on sending it out sometime after the monday morning meeting once I've managed to caffinate. If there are any questions/concerns/missing items/clarity, etc. Just highlight it and we'll go from there. | 22:58 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!