Update or removal from documentation needed for "install.sh" installer script

[rampa3]

LocalAI version:

v3.9.0 (aadec0b)

Environment, CPU architecture, OS, and Version:

--- (Does not pertain to particular install of LocalAI)

Describe the bug

The install.sh script listed as an installation method in the documentation at URL https://localai.io/installation/linux/ seems to produce broken/misconfigured installations of LocalAI according to users who asked for troubleshooting help in LocalAI Discord.

To Reproduce

--- (Needs testing... Did not had time to set up a VM to test it in yet.)

Expected behavior

Script should properly install either Docker or binary copy of LocalAI.

Logs

--- (Not available - users I helped with troubleshooting did not provide console output of the script)

Additional context

Creating this issue after helping to troubleshoot two broken fresh installs using install.sh back-to-back in #help channel on LocalAI Discord. At this point, I think the script either will need an update to run properly for new versions of LocalAI, or should be hidden from the documentation's "Installation" section.

Personal testing will be conducted in a VM when I have free moment to conduct it. If I will identify the issue at that point, I will attempt to write a fix.

[mudler]

@rampa3 do we have logs or any traces of what happens, but in any case yes I agree better to disable it until it's fixed

[rampa3]

@rampa3 do we have logs or any traces of what happens, but in any case yes I agree better to disable it until it's fixed

Sadly we don't. I tried to ask for more info but did not get it.

What I know though is what the users complained about:

• script installed install of LocalAI did not worked properly for ROCm for some reason while Docker did - this one I did not got info for sadly
• the other one now looking back is a user confusion - they mixed installer script's variables with LocalAI variables, because they looked into the script for how to set LocalAI environment - maybe we should add a message that tells users how to configure the environment variables for deployments that needs this? (Just posted a follow up mention to the user, as I got a bit confused there too...)

That outlines two things to investigate/act upon at least as I see it for now:

• Check if ROCm works properly on script installed binary install to rule out user error. (Sadly don't have AMD card to test it by myself...)
• Look into adding a message with a link to a detailed list of all current configuration environment variables at the end of the install script.

I hope I have not mixed up or forgot anything to list here.