I am still learning pacman while Eric is probably ready to scream, "what does this have to do with 66?"

Ok, I will explain. I have been trying to come up with a procedure (possibly a script in the making) that would reliably transform any arch installation into an Obarun system, emphasis on reliable, as I am not much of a script writer so when something fails I can't see how the routine would be interrupted and exit with meaningful output, let alone further continue where it was left off. Maybe I should read obarun-install in more detail.

In one attempt I ended up with a systam that no matter WHAT I tried I couldn't make root tree be initialized at next boot. One of the errors I discovered on boot was it couldn't read /var/lib/random-seed, then I discovered that arch had 0600 on the same file that I somehow had ended up with 0644?? Then I inquired on where this file would be coming from, it must belong to some pkg, right? Neither pacman -Qo or -Fo revealed who the owner is. Also I noticed on arch based distros this is a 4k file, on debian-based /var/lib/urandom/random-seed is half a kilobyte. ?? If booting just reads this file and it was at read only stage of booting even 0400 shouldn't have been a problem, right?

Somehow I overcame the problem of reading /random-seed but even then something was throwing the booting off, while I was waiting for console to come on I would skip to tty12, but instead if being logged in as a user I would get this emergency script of entering the root pw to go in and fix things. If I tried to exit this stage even F12 would be locked and only a hard crash/reboot would get me out then. If I could manage to manually initialize root I would get tty1 and tty2 and everything would be ok, till next reboot. I lost my secure feeling that no matter what I can go in tty12 and fix things.

I suspect what the issue was is the autologin mechanism set up in arch-live that I couldn't disable before or after the transition. Maybe it is some init-script thing that is systemd controlled and after removing it all it stays hidden somewhere and not functional, but throws s6 booting off.

Doing a true arch installation with pacstrap is no problem at all, in 5' it is switched into Obarun and runs flawlessly.

The question remains "where does random-seed come from" and why am I having problems finding out where.
I can barely theoretically understand why it is used and why it is important to have it, but still, does it belong in filesystem? ???
4 days later
fungalnet wroteI am still learning pacman while Eric is probably ready to scream, "what does this have to do with 66?"

Ok, I will explain. I have been trying to come up with a procedure (possibly a script in the making) that would reliably transform any arch installation into an Obarun system, emphasis on reliable, as I am not much of a script writer so when something fails I can't see how the routine would be interrupted and exit with meaningful output, let alone further continue where it was left off. Maybe I should read obarun-install in more detail.

In one attempt I ended up with a systam that no matter WHAT I tried I couldn't make root tree be initialized at next boot. One of the errors I discovered on boot was it couldn't read /var/lib/random-seed, then I discovered that arch had 0600 on the same file that I somehow had ended up with 0644?? Then I inquired on where this file would be coming from, it must belong to some pkg, right? Neither pacman -Qo or -Fo revealed who the owner is. Also I noticed on arch based distros this is a 4k file, on debian-based /var/lib/urandom/random-seed is half a kilobyte. ??
Well, to know what do a thing just read the scripts :lol:.
Seriously, a little error exist on this service but anyway the /var/lib/random-seed file is created by the rwfs-random service and that's why you cannot see it into a packages. this file is created in function of the /proc/sys/kernel/random/poolsize file. This file is given by the kernel and a dd operation is make with this value to create the /var/lib/random-seed. So the size will not depends (should not depends) of the distro but it depends of the kernel.
" 0600 on the same file that I somehow had ended up with 0644??" weird, i have 600 on my machine, need to check this...
If booting just reads this file and it was at read only stage of booting even 0400 shouldn't have been a problem, right?
you can always read a filesystem even if the filesystem is a read-only one, so right.
Somehow I overcame the problem of reading /random-seed but even then something was throwing the booting off, while I was waiting for console to come on I would skip to tty12, but instead if being logged in as a user I would get this emergency script of entering the root pw to go in and fix things. If I tried to exit this stage even F12 would be locked and only a hard crash/reboot would get me out then. If I could manage to manually initialize root I would get tty1 and tty2 and everything would be ok, till next reboot. I lost my secure feeling that no matter what I can go in tty12 and fix things.
this is a classic sulogin, just press CTRL+D to exit from this one. This trick should be displayed at the start of the sulogin. Also CTRL+ALT+DEL should reboot your machine.
The new 66 release change a lot all this behaviour. The shutdown process is completely rewritten.
I suspect what the issue was is the autologin mechanism set up in arch-live that I couldn't disable before or after the transition. Maybe it is some init-script thing that is systemd controlled and after removing it all it stays hidden somewhere and not functional, but throws s6 booting off.
nope it come with boot-66serv, again this will change. But you can control this behaviour by editing your /etc/stage2 file and comment the last line marked ${ISHELL}. Now in case of crash, the sulogin will not opened and anything will be opened. If you want a sh interactive opened just change the litteral ${ISHELL} variable by: sh -i.
The question remains "where does random-seed come from" and why am I having problems finding out where.
I can barely theoretically understand why it is used and why it is important to have it, but still, does it belong in filesystem? ???
generally speaking the answer is not because this file need to reflect your system and so it written by the system for the system and so cannot be installed by a package. But in practical way the file is a part of the filesystem because without it you goes on trouble.
Yes, after some more research it seems that pretty much all systems during the rwfs-random boot stage use dd if=urandom count=xx of=/var/lib/random-seed (and some place it in /var/lib/misc/random-seed) and some make it 4MB some 500KB. I wonder if games that use random mechanisms, such as board/dice games, rely on this file.

Anyway, I run into it due to the error and for encyclopedic interest I was trying to find out where it came from.

Powered by Obarun