| caribou | blackboxsw: thanks for the precision | 07:37 |
|---|---|---|
| shind | is there a way to use "ca_cert" trusted but from an existing file instead of having an inline certificate? | 09:39 |
| shind | the scenario is to use "bootcmd" and call "aws secretsmanager get-secret-value", parse it, save it as a crt file and then install it using the ca_cert. | 09:39 |
| shind | i know i can just use the built-in linux command of updating the ca but i wanted to check it cloud_init supports reading from files | 09:39 |
| meena | sh… | 10:40 |
| surfzoid | hi | 16:07 |
| surfzoid | I'm trying to build a deb of my qt project on launchpad, locally on m RPI400 it compil perfect, but sound like launchpad don't manage well rpath : https://launchpadlibrarian.net/657995979/buildlog_ubuntu-kinetic-amd64.qtvsplayer_1.0.39-0~202303271523~ubuntu22.10.1_BUILDING.txt.gz i'm a pure noob with deb and launchpad, is there a easy workaround in the deb files? | 16:07 |
| minimal | surfzoid: that doesn't appear to have anything to do with cloud-init | 16:20 |
| surfzoid | minimal: i see i see it too late sorry | 16:21 |
| sbraz | hi, i just filed this bug downstream for fedora 37 but i thought maybe someone here had an idea of how to avoid this kind of problems with NICs taking a long time to come up → https://bugzilla.redhat.com/show_bug.cgi?id=2182173 | 18:46 |
| -ubottu:#cloud-init- bugzilla.redhat.com bug 2182173 in Fedora "cloud-init-local.service fails with 'Unable to find a system nic' with some Mellanox NICs" [Medium, New] | 18:46 | |
| meena | sbraz: which Mellanox NICs? what DataSource? | 18:47 |
| sbraz | meena: connectx-4/5, NIC details are in the bug report, datasource is a configdrive | 18:48 |
| sbraz | meena: but i don't think the issue is specific to this nic model, it's more of a "how to ask cloud-init to start when the NICs are actually available" question, and i don't really know | 18:50 |
| minimal | seems like your change to re-add the udev-settle is the right way | 18:55 |
| sbraz | minimal: i just don't understand why i'm not hitting this issue with e.g. debian 11 which doesn't use udev settle | 18:55 |
| sbraz | gotta run, i'll read all this later/tomorrow, thanks for the feedback :) | 18:55 |
| minimal | (or equivalent that isn't deprecated) | 18:55 |
| minimal | sbraz: when comparing Fedora and Debian I assume they are both not using the same kernel version, same versions of other software - i.e. too many variables to properly compare | 18:57 |
| waldi | sbraz: AFAIK, fedora shuffles around when they run stages | 19:10 |
| waldi | early boot is not stable and cloud-init got some problems there anyway | 19:11 |
| sbraz | waldi: what do you mean exactly? | 23:19 |
| sbraz | minimal: that's the thing, i don't know that there is a non-deprecated equivalent | 23:19 |
| minimal | sbraz: I don't use systemd so can't suggest any alternative to udevadm settle | 23:42 |
| minimal | I'm guessing different MLX cards take different amounts of time to initialise (e.g. some of them are IB only, some support IB and ethernet). It's been a while since I've played with MLX kit | 23:43 |
| minimal | I just remember some cards using mlx4 driver and others using mlx5 lol | 23:44 |
| sbraz | minimal: it really is a matter of seconds, the connectx-3 using mlx4 is fine for instance; maybe i could try to move the driver to the initramfs to speed up its loading time, i'll check | 23:54 |
| minimal | sbraz: that might reduce the likelyhood of the race condition but it's unlikely to guaranteed its elimination | 23:56 |
| minimal | I was using MLX cards with cloud-init and don't remember this sort of problem....but that was 5+ years ago lol | 23:57 |
| sbraz | yeah the correct fix is to wait for nics to be ready, and that's it, no hack or anything | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!