| smoser | felipe1982: if you have an older version of cloud-init, you may not have the cloud-init-output.log enabled. and it will just go to wherever the init system output goes (probably /dev/console) | 03:08 |
|---|---|---|
| smoser | but, yeah, you probalby can run that as blackboxsw pointed out | 03:08 |
| felipe1982 | amazon linux 2 is quite new. I would have assumed it was latest [enough] | 03:24 |
| felipe1982 | what version would I need for *boothook* to write to /var/log/cloud-init-output.log | 03:25 |
| Odd_Bloke | Just realised that I did all of my testing for the Oracle secondary VNICs branch with a local change to prefer data source network config over initramfs config. | 14:17 |
| Odd_Bloke | So a follow up MP will be coming Real Soon Now. | 14:17 |
| Odd_Bloke | rharper: blackboxsw: https://code.launchpad.net/~daniel-thewatkins/cloud-init/+git/cloud-init/+merge/371403 <-- actually enable Oracle secondary VNICs (instead of just generating config which would enable them and throwing it away) | 14:38 |
| smoser | :q | 15:24 |
| smoser | oops | 15:24 |
| Odd_Bloke | rharper: On https://bugs.launchpad.net/cloud-init/+bug/1799301, I think the problem is that it _does_ use sysconfig and we don't detect that we should be using sysconfig, right? Or are you wondering if the Fedora comment is actually a distinct bug? | 15:40 |
| ubot5 | Ubuntu bug 1799301 in cloud-init "SUSE sysconfig renderer enablement incomplete" [Medium,Incomplete] | 15:40 |
| rharper | Odd_Bloke: yes; but it seems to be related to where the additional files we check are located | 15:40 |
| rharper | 'etc/sysconfig/network-scripts/network-functions', | 15:41 |
| rharper | 'etc/sysconfig/network-scripts/ifdown-eth' | 15:41 |
| rharper | we look for those; I wondered why Fedora wouldn't have those if they're still using sysconfig | 15:41 |
| rharper | w.r.t OpenSuSE; they have different paths | 15:41 |
| rharper | at least in some of the releases; hence the bug | 15:42 |
| blackboxsw | per my ubuntu-drivers branch I think there is an option to provide debconf-set-selections instead of cloud-init knowing to write out /etc/default/linux-modules-nvidia. I'm exploring that option now since we already do debconf-set-selections for byobu, ca-certs and grub | 17:09 |
| rharper | blackboxsw: ok; writing out the line in the conf is going to be faster/cheaper than invoking debconf ; if the config value or what needs emitting is expected to change ; that does abstract cloud-init from those sorts of changes; | 17:37 |
| blackboxsw | rharper: Odd_Bloke smoser: for those interested, the 2nd commit on my branch updates cloud-init to just use debconf-set-selections, making it also a much smaller branch. (assuming that is the right selection value. I like abstracting cloud-init from the config implementation details if possible, in case they change. Even if it is a little slower for the corner-case deployment where cloud-init user is on gpgpu. | 18:00 |
| blackboxsw | https://code.launchpad.net/~chad.smith/cloud-init/+git/cloud-init/+merge/371369 | 18:00 |
| blackboxsw | so if we like the 2nd commit more than the first. I can rebase and squash of we get confirmation that debconf-set-selection is valid | 18:00 |
| blackboxsw | *if we get confirmation* | 18:03 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!