bc gawk grep procps sed ncurses-bin wget cron whiptail man-db groff-base rsyslog
file less sudo tar util-linux openrc vim curl rsync sqlite3 ? groff-base
causes uptime, htop, top to fail
env & last checked
ish-Debian missing 24-02-19
Not sure what happens, seems to create at least 3 /bin/sh /usr/local/bin/aok -e off processes, ending up non responsive in most cases new ssh sessions can be created. existing sessions tends to survive as long as they didnt generate any output whilst the -e off took place
htop from A 3.14 apk add lua5.3
Verify the new grep check works as intended
Set it with a reasonable PATH
- option for logging console to pts/0
- option for lsunch cmd: default, aok default, custom
- should bat lvl be displayed in colors yes/no
- bat levels default / custom
- bat lvl colors
Alpine/usr_local_bin/apk-find-pkg.sh it is now apk-find-pkg
Investigate if anything can be simplified / needs to be changed given that /run is now clean at boot time for iSH-AOK, and asuming dynamic_login is the Launch cmd, this is also the case for regular iSH hm perhaps this cleanup of /run should be the first inittab task instead?
an improvement would be to make it a generic tool, then just keep something like .shellchecker in the top dir of a project, defining exclusions
investigate and if confirmed, try to find latest version that can be used, and add it to set of custom apks in Alpine/setup_alpine.sh
This actually works much better in Debian than in Alpine, since in Alpine as of now only auto-login as root works. agetty fails to change ownership of /dev/pts/0 on Alpine
1 Add this as Launch cmd to avoid harmless but annoying error msg everytime
ish is started and offensive login BEFORE init is run, but still ensure
/dev/pts/0 is bound
/bin/sleep infinity
2 Replace console with this alternate content in /usr/local/sbin/fix_dev if you use it via inittab, otherwise run it in a shell as root, in order to ensure anything can print to /dev/console, without being restricted when agetty locks down /dev/pts/0 With this normal bootup console output can be seen!
#rm -f /dev/console && mknod -m 666 /dev/console c 5 1
rm -f /dev/console && mknod -m 222 /dev/console c 136 0
3a Add this towards end of /etc/inittab if you want a login prompt for console screen
pts0::respawn:/sbin/agetty pts/0 linux
3b Alternatively use this if you want to use -a to login as a user without prompting for password, be aware that in this case logout / exit will instantly automatically log you back in again 🙂 You will have to use shutdown to terminate the iSH app
pts0::respawn:/sbin/agetty -a root pts/0 linux
IMPORTANT UPDATE: Please be aware that in Alpine you cant use pts0 as an inittab identifier for whatever reason, despite it being no longer than 4 chars, in such cases labeling it as tty1 works and will give you a prompt. On Debian pts0 works, and makes more sense since it hints what device this is using
runlevel default should, do, can something else be done if openrc is not used?
This could be used from /etc/profile for bash/ash and from /etc/zprofile for zsh
It needs to be cheap enough to not noticeably delay future login shells Here are my current ideas
- Simplest, but assumes runbg is an active service... Check if /run/openrc/options/runbg/pidfile exists and is newer than /run/runlevel
- If AOK_HOSTNAME_SUFFIX="Y" and ish-AOK this is quite cheap and quick
while ! hostname | grep -q '\-aok' ; do
- Uses ps ax, so has a crash risk, probably not ideal
while ! ps ax | grep [i]nit | grep -q '\[2\]'; do
rest of this code block
echo "waiting for bootup to complete"
sleep 2
done
When the deploy starts its pretty clear what hostfs_is_alpine
and
destfs_is_alpine
is refering to. However when the destfs boots up and
does a large part of the deploy itself, shouldnt it be the host?
Perhaps it should be seen as a chroot thing. If something is working chrooted on a buildhot it would make most sense to see that as a destfs in this context, but if the same deploy steps happens on the deploy target, running the deploy as its primary env it would ssem to be the hostfs. More clarity about how to refer to different roles needs to be found.
it seems a respawn process only is run once
When used as a Launch cmd, and autologins to a zsh user the console session is logged out after a copple of minutes - investigate
- ensure NavKey.md has correct paths
since it is about to become more usefull, i should update it to ensure it is in line with the debian image when it comes to what is installed out of the door