#it was observed that that setting timedatectl set-timezone <zone/subzone> does not persist on RMPP r

1 messages · Page 1 of 1 (latest)

kind island
#

Related to this point #rm-pro message

#

So, wanted to double-check, unmount the overlay on /etc it means running this command?

mount -o remount,rw /etc
cedar plume
#

overlays are already read-writable. That will do nothing. You need to unmount it.

kind island
cedar plume
kind island
#

Yes, worked. Thank you 🙂

loud crown
#

something to be aware of ... umount -R /etc will un-mount two different mount points: /etc is an overlay, and /etc/dropbear is a "bind-mount". if you wanted to set things back to normal without rebooting, you would need to run three commands:

#

mount -o rw,relatime,lowerdir=/etc,upperdir=/var/volatile/etc,workdir=/var/volatile/.etc-work -t overlay overlay /etc

mount -o bind /home root/.dropbear /etc/dropbear```
#

IN THAT EXACT ORDER

#

you can get the second command's -o option's value by running mount | grep /etc before un-mounting anything

cedar plume
#

@loud crown a PR to update the FAQ would be appreciated

loud crown
#

(i spent about half an hour on this yesterday)

#

i was actually looking at that. what language are .rst files? i've never seen that notation anywhere other than this one repo

loud crown
#

i had to add some custom scripting to add the git info at the bottom of the ToC pane and the pages, but that was fairly easy, and to be honest i enjoyed figuring it out.

cedar plume
#

You can get sphinx to parse markdown, but restructured text is way more powerful

loud crown
#

is there a way to see a "live preview" while working on the files? with mdbook you can run mdbook serve and it runs a live web server on the local workstation, then opens a browser to http://127.0.0.1:3000 or whatever the port is, and then as you save changes to the source files it re-runs the engine and makes the browser reload automatically

loud crown
#

here i was about to suggest adding that info to the repo's README.md file, only to discover it's there already ... today's going to be one of those days i guess

#

might be worth adding specific directions for how to install sphinx and the other pre-requisites, including which distro those package names are meant for (debian and redhat use different names, plus homebrew on macOS has its own slightly different names as well - i'll see if i can come up with the right brew install command to install it all

cedar plume
#

I'd accept a PR to add some extra information on that though

#

That said, you don't need to install sphinx, the makefile installs it for you in the venv it uses

loud crown
#

i'm assuming this list is for something debian-flavoured ... build-essential is a meta-package that includes over 100 other packages. do you happen to know which tools are actually needed? i'm guessing gcc and g++ are not needed ...

cedar plume
#

I believe gcc may have been required for some of the python dependencies

#

I can't remember

loud crown
#

okay ... that just means it won't be as easy as i had thought. i'll have to create a VM running macOS, install only homebrew plus the package names i've already identified, then run make and see what screams. which means it won't be done in half an hour. i'll go ahead and create a PR with just the mount commands then.

#

actually ... i'm looking at the Makefile, the dev: target appears to have --port=0 ... i'm wondering if i change that to a known value and remove the --open-browser option, could i run this in a debian 12 docker container with that port mapped, and just manually point a browser at it?

cedar plume
#

You could

#

docker container would probably be the easiest

loud crown
#

already figuring out those details ... once i've got something coherent i'll make a different PR, adding a totally new directory to the repo, with the relevant files and a README there

#

what distro was that list written for anyway?

cedar plume
loud crown
#

i just noticed one other thing: umount -l /etc should be umount -R /etc ... otherwise it will remove the root filesystem's /etc entry pointing to the overlay (so the root filesystem's real /etc directory will be "visible") but not actually un-mount the overlay filesystem itself, because /etc/dropbear (a bind mount) will still be mounted. watch for another PR in a few minutes.

cedar plume
#

You don't want to unmount dropbear while SSHing into the device, you can break things

#

So make sure to handle that

loud crown
#

it's just a bind-mount that makes /etc/dropbear contain the contents of /home/root/.dropbear ... i'll add a note about this, need to figure out how to do this markup language's version of a "note", which markdown uses > for

cedar plume
#

Probably a lazy umount of /etc/dropbear followed by a umount of /etc

#

and then mounting it all back

#

as dropbear would be accessing that folder

loud crown
#

i've been using -R, when the command finishes both filesystems are un-mounted

cedar plume
#

Oh interesting

#

It must be translating the file handle to the actual location underneath the bind mount

loud crown
#

the tablet mounts the overlay filesystem as /etc, then bind-mounts /home/root/.dropbear as /etc/dropbear ... they have to be mounted in this order, because mounting /etc after mounting /etc/dropbear would "hide" the bind-mounted /etc/dropbear directory and processes would see whatever /etc/dropbear happens to contains within the overlay filesystem (which would be a "view" of what it contains on the read-only / filesytem, unless somebody had given it new contents)

#

the real files are in the /home/root/.dropbear/ directory, this is how the rMPP is keeping the same SSH host keys across OS upgrades.

cedar plume
#

I'm glad they resolved that on the rMPP. I haven't looked at the latest rM1/2 update files to see if they've changed those to also do it

loud crown
#

so it looks like the syntax for an indented text block is to just indent the lines with 4 spaces, easy enough ... can a code-block be included within the indented block? (i don't have make dev working in a container yet, it should be but colima isn't making the exported port available from the macOS host for some reason. probably some new apple security thing, grrrrr)

#

just created another PR ... fair warning, i'm guessing that the "code block within an indented paragraph" thing is going to be formatted correctly. you may want to preview it yourself and fix it if it doesn't look right.

#

that PR's build failed, but github doesn't want to show me any useful error messages

cedar plume
#

I believe there are some extra rules on the indentation where it gets annoyed otherwise.

#

Afk for a bit though