Split installation onto two "drives"

I can possibly get my hands on some LE suitable devices (from employer). These

have a pair of small-ish flash devices soldered down. I am hoping that I can

fit the "system" on the first (512MB) and some few add-ons on the second.

There will be no music, video, etc. files stored on the devices so limiting the

add-ons will put an upper bound on how much of the second drive I need.

The stock installer (I'm still playing with 9.2.6) appears to want a single

drive that it can split into two partitions (sda1/2). I am hoping that I

can let it install onto spinning rust and then "move" the two partitions'

contents onto these two flash devices -- tweeking syslinux.cfg and their

associated disk labels, accordingly.

Does anyone see any problems with this approach?

I want to get it working (on two "drives" that have been deliberately

limited to 512MB in size) before I ask my employer for any hardware...

Thanks!

Quote from lrusak

You would have to do it manually as the installer doesn't let you do such a thing.

Yes, the installer expects both partitions to reside on the same $INSTALL_DEVICE -- that was the reason for my post.

Quote from lrusak

You could either manually create the partitions and installations or you could install to one drive, then format the other driver and update the partition UUID in the bootloader.

I opted to build a "one drive" installation on a ~500G drive. I added all of the appropriate ADD-ONS and configured it for use. Then, manually create the two (small) FLASH drives and copied the contents of the two rust partitions onto the two FLASH drives. This seems to work...

If I find some "spare time" (ha!), I may revise the installer to provide this option.

But, my next goal is to see how much further I can reduce the "disk" requirements...

Quote from Da Flex

You are wasting your time by adding limitations to LE. Storage space is cheap these days.

When you are designing appliances, every byte "wasted" is cost -- profit left on the table. You're assuming ALL you want a device to do is be a media player... how 1970's! A box for this, a box for that, a cable here, a cable there, a wall-wart here, another there... I'm going to embed these in the wall -- no unsightly boxes littering the room, no silly wall-warts poking out of electric outlets, no "remotes" to leave lying around the room, etc. Just a cable to the back of the TV. Something more appropriate for the 21st century.

And, of course, it's my time. Are you suggesting I not share any changes I make to the installer script as those simply "reduce the storage requirements"? Obviously, it will save me a fair bit of time if I don't have to craft a robust script and can just settle for something that suits my needs, and my needs alone...

Quote from Da Flex

In case you're thinking, OS and add-on's can be separated for a long time, you're wrong. As an example, current Widevine changes require OS changes to keep it working on ARM. All your byte-per-byte optimizations can be a waste of time then. Things change quick in nirvana world.

I'm just using LibreElec to illustrate "proof of concept" -- so I can demonstrate functionality without "little boxes", wall warts, remotes, etc.. My little devices will devolve into little more than video spigots moving ethernet packets onto a COTS display -- all of the content processing is done remotely (so every "display" can benefit from the same capabilities at the head end). E.g., LibreElec doesn't support PIP -- yet I can demo that, today!

Quote from Da Flex

Ah, now I get the idea, "demonstrate" was the keyword to me. If you make it, don't forget the German market.

Yes. People are amazingly unimaginative; you can tell them what you intend to do and they typically won't be able to make the leap to seeing it (in their mind): "What do you mean, 'no little boxes'? 'No remotes'?"

But, if you hack something together that approximates your end result and then show this to them -- working -- they suddenly "get it". They seem to need to be able to touch a real thing and see how it works... If I implement the functionality in a PC -- regardless of how small it is -- they'll see a PC sitting by the TV. If I can show them a *tiny* bit of electronics, hiding in the wall, and tell them how much smaller/cheaper it can still be made...

As to markets, I'm just an engineer; I care very little about how or where it's sold. I just "solve problems". Note that the "in the wall" box has no user interface so it cares nothing about i18n/l10n!