[16:13] <gjolly> Hello there, could you have a look at this bug and at my last comment? https://bugs.launchpad.net/livecd-rootfs/+bug/1968873. Basically it seems that cloud-init does not support/use sshd_config.d. Do you think it is a cloud-init bug?
[16:13] <ubottu> Launchpad bug 1968873 in livecd-rootfs "Regression: images ship with modified configuration file" [High, In Progress]
[16:17] <gjolly> I feel like cloud-init should write its config to sshd_config.d/99-cloud-init.conf or something like that rather than appending to the main config file.
[17:56] <rbasak> gjolly: thanks - I commented in the bug.
[17:56] <rbasak> blackboxsw: ^
[19:42] <Odd_Bloke> Glad to finally see that being addressed!
[20:09] <kovacs-andras> Dear Team and friendly folks around here, I have a few Ubuntu servers (20.04 mostly) where the system partition is luks2 encrypted. I could open it on boot time trough a tty, ttyS and ssh(dropbear). SSH works just fine I call cryptroot-unlock explicitely.
[20:09] <kovacs-andras> And previously both tty and ttyS worked too.
[20:09] <kovacs-andras> But now, either tty or ttyS works only.
[20:10] <kovacs-andras> I can't figure out from the changelogs what is different now.
[20:11] <kovacs-andras> I put into my /etc/default/grub this: GRUB_CMDLINE_LINUX_DEFAULT="console=tty0 console=ttyS1,115200n8" and it worked fine a (months ago).
[20:14] <kovacs-andras> Now cryptroot-unlock only asks for my input on the "last" console in this variable. So I can change the order and always only one of them works.
[20:15] <kovacs-andras> tty numbers really doesn't matter for some reason. So in case tty* is the last in the row, it will ask for my pass on the tty. If ttyS1 is the last, it will ask my pw on that one.
[20:20] <kovacs-andras> Do you have any recommendations on what should I change to make all my consoles work again? Should I make a hook/script now?
[20:35] <sarnold> kovacs-andras: this is outside my usual experience.. but I have a vague memory of plymouth being used to do password prompting most of the time; I wonder if this is a matter of plymouth config, rather than the right console= lines?
[20:36] <sarnold> kovacs-andras: the kernel docs mention the last console= is used when processes open /dev/console https://www.kernel.org/doc/html/v4.13/admin-guide/serial-console.html -- that sounds very much like what you've got now
[20:43] <kovacs-andras> thanks!
=== NotEickmeyer is now known as Eickmeyer
[21:04] <kovacs-andras> I used ssh(dropbear) to enter into my running initramfs. Both tty0 and ttyS1 are working fine. I can read-write them manually with cat. But cryptroot/cryptsetup runs only on one of them.
[21:15] <kovacs-andras> Many thanks sarnold! Now I understand the last console listed appears as the primary console. But I really don't understand how it could work previously... and I still don't know how to make it work again.
[21:17] <sarnold> kovacs-andras: yeah, I'm not sure it would work previously either :( that's where I started grasping for eg plymouth, thinking it might have done something different as part of its IO multiplexing..
[21:54] <kovacs-andras> Previously it worked for me. Now I'm thinking about a dirty hack like this to execute unlock on the "secondary" terminal: sh -i <>/dev/ttysomething 1>&0 2>&1 /bin/cryptroot-unlock