Porteus USB live install

[M. Eerie]

I have downloaded the last two commits (porteus-{usb,net}-install.sh). This is what I got:

NET INSTALL PASTEBIN

USB INSTALL PASTEBIN

Besides, porteus-mirror-selection.sh continues to operate the same.

Almost forgot... My pendrive was indeed partitioned, but the 1st partition is of unknown gender and the rest is free space.

Minghia...

[robang74]

ABOUT NET INSTALL & MIRROR SELECTION

• ERROR: folder '/tmp/test/moonwalker' does not exist, abort!

Cleaning is something useful when we like to start from scratch. So it was not need in your case, hence error. Despite this, you followed the instructions to see what was going to happen. Error. Create folder, clean a void folder. done. Moreover, as you can see "--clean" is not present (documented) into usage, an probably it is better that I restrict to developers (using DEVEL=1 as enviroment variable, which is not fully support because sudo does not use -E option, in v0.3.1).

The shortlink 2csl7ae3 says that you have tested the v0.3.1/ but to obtain a different result you should have used a different version. Currently there is none above v0.3.1, hence main but main is not supported at the moment by the script. I need to fix this lack using an environment variable: DEVEL=1 bash -i <($url_to_main_head). So how to test the last script in the repository?

• https://github.com/robang74/porteus-usb ... /main/TODO

In this file that usually only others DEV looks into, there is another link: tinyurl.com/2bn89qqo (RISKY!!)

The first question is about downloading a version from tags and here we have same issue: main is not included.

Code: Select all

roberto@x280[3]:/tmp/test$ bash -i <(wget -qO- tinyurl.com/2bn89qqo)

-> pwd: /tmp/test/moonwalker

###############################################

This is an on-demand from remote running script

###############################################

WARNING: script suggests 'v0.3.1' as last tagged, check for updates? [Y/n] 

Using DEVEL=1, things changes and the script on the HEAD of the main branch, show main as tag version

Code: Select all

roberto@x280[3]:/tmp/test$ DEVEL=1 bash -i <(wget -qO- tinyurl.com/2bn89qqo)

download path: /tmp/test/moonwalker
workingd path: /proc/3035746/fd

-> pwd: /tmp/test/moonwalker

###############################################

This is an on-demand from remote running script

###############################################

-> tag: main (DEVEL: 1)

WARNING: no any mirror selected, using 'https://mirrors.dotsrc.org/porteus'

Do you want to check for the fastest mirror available? [Y/n] 

Another issue is about asking for selecting the best mirror WHATEVER we are going to download the ISO or not. It is a UI/UX mistake but nice/easy for testing that feature which at the moment is more important than being strinct on UI/UX logic.

The correct answer would be - in general 'n' unless we are going to download the ISO - but the default mirror in v0.3.1 is affected by a bug. Therefore 'y' is mandatory hoping that the selected one is NOT the same as default which triggers the bug.

IN SHORT

Use DEVEL=1 to run 'main' head scripts or wait for the coming soon v0.3.2 tagged version.

• https://github.com/robang74/porteus-usb ... s/main.zip

The always-updated main.zip archive can be downloaded at any time visiting the github page and using the green button to download it.

Once downloaded the main.zip, you can use the script in local with `bash script-name.sh` or simulating a remote on-demand with `bash -i <(cat bash script-name.sh)`

Code: Select all

mkdir test; cd test; wget -qO- https://github.com/robang74/porteus-usb-installer/archive/refs/heads/main.tar.gz | tar xvz; cd porteus-usb-installer-main

[robang74]

This mirror has zero size file sha256sums.txt - for example - this is a condition that the script is unable to deal with, yet

Code: Select all

Welcome to ftp.nluug.nl
The large public mirror server for open source projects hosted by NLUUG and Surfnet

/os/Linux/distr/porteus/x86_64/current/
File Name  ↓ 	File Size  ↓ 	Date  ↓ 
Parent directory/	-	-
bundles/	-	2024-Aug-19 09:08
kernel/	-	2025-Feb-02 19:09
language/	-	2023-Sep-25 12:28
modules/	-	2025-Mar-24 09:00
updates/	-	2025-Feb-02 22:09
Porteus-CINNAMON-v5.01-x86_64.iso	436.0 MiB	2024-May-20 07:55
Porteus-GNOME-v5.01-x86_64.iso	436.0 MiB	2024-May-20 07:56
Porteus-KDE-v5.01-x86_64.iso	495.0 MiB	2024-May-20 07:54
Porteus-LXDE-v5.01-x86_64.iso	421.0 MiB	2024-May-20 07:54
Porteus-LXQT-v5.01-x86_64.iso	435.0 MiB	2024-May-20 07:54
Porteus-MATE-v5.01-x86_64.iso	428.0 MiB	2024-May-20 07:55
Porteus-OPENBOX-v5.01-x86_64.iso	417.0 MiB	2024-May-20 07:55
Porteus-XFCE-v5.01-x86_64.iso	426.0 MiB	2024-May-20 07:56
sha256sums.txt	0 B	2025-Mar-24 09:01

• Source: https://ftp.nluug.nl/os/Linux/distr/por ... 4/current/

But soon it will deal with a whole category of anomalies like this one above, in this way.

Code: Select all

roberto@x280[4]:/tmp/usbtest$ DEVEL=0 bash -i moonwalker/porteus-net-install.sh

-> pwd: /tmp/usbtest/moonwalker

WARNING: script suggests 'v0.3.2' as last tagged, check for updates? [Y/n] n

ERROR: no ISO name or its checksum found, abort!
Catalog filename: sha256sums.txt
---------------- content start -----------------
----------------- content end ------------------

While the mirror selection will refuse to accept those mirrors with missing information.

WARNING WARNING WARNING

Without that file, the ISO file and their contents can be compromised by everyone that has the write permission on that server OR earn it in any possible (and also illegal) way!

Compromising ISO files, means that it is possible to compromise also the computers. Considering BIOS, firmware and MBR combined together, it can be a permanent compromission.

[list]https://robang74.github.io/chatbo ... -fw[/list]

Why do PCs still have a BIOS? (above) and which impact can have a firmware as BIOS?

[list]https://robang74.github.io/robert ... tml[/list]

A this point, I am feeling safer in downloading 438MB from p0rnhub rather than from such a mirror...

[vinnie]

I think that if people posting and commenting on technical forum would stick to "facts" - all these "emotional psychodramas" will end soon.

Let me reply just one times, I recognize that my message is offtopic and I apologize to others and to you as well, since the topic is yours.
A distribution needs all kinds of participants, not just ultra-technical users who are needed but do not form a community by themselves.
For example you can program everything but you will need translators if you want to internationalize your project and it is not certain that they are all programmers.
No one can claim to know everything but blaming others for alleged lacks eventually leads people to not want to participate in a project.
So what we have in common is perhaps also the ability to interact with others respectfully.

Bah, disse un saggio - non basta una vita per conoscere se stessi, eppure c'è gente che giudica.

On this I agree with you, there are very presumptuous people out there

Off topic closed

[robang74]

A distribution needs all kinds of participants, not just ultra-technical users who are needed but do not form a community by themselves.

s/a distribution/a society/

Code: Select all

but do not form a community by themselves. 

WRONG! -- WE are able to start-up a civilisation, even when left alone in the wild.

Think about twice next time a group decide to follow ONE OF US down from the branches of a three.

It is also incredibly arrogant thinking that WE are few and cannot do something about ourselves, alone. Arrogant as much as ants would think that they are important just because humans avoid to step and walk on them. Good humans, don't do that. Unless, it is necessary (cit.).

@vinnie - Gli italiani nella loro banalità di comunisti, pensano che nasciamo tutti uguali, quando invece siamo tutti uguali in termini di diritti umani, ma siamo tutti diversi già a partire dal colore degli occhi e dei capelli, e curiosamente però sono fra quelli che fanno la gara a chi c'è l'ha più lungo, proprio perché cresciuti in questa assurda ideologia. Curioso come i bias culturali conducano ad allucinazione collettive. Quindi anche tu sei pregato di NON intervenire ne off-topic e ne on-topic riguardo ai miei post. Grazie della comprensione, R-

M. Eerie - Screenshots are overkilling, but `sudo fdisk -l /dev/sdx | grep -e /dev/sdx` -- it brings the necessary informations without exposing device identifications ids.

Code: Select all

+:git-shell:main:porteus-usb-installer> sudo fdisk -l /dev/sda | grep /dev/sda
[sudo] password for roberto: 
Disk /dev/sda: 58.59 GiB, 62914560000 bytes, 122880000 sectors
/dev/sda1  *     2048     34815     32768   16M 17 Hidden HPFS/NTFS
/dev/sda2       34816 122879999 122845184 58.6G 83 Linux

This is HOW a EXT4 persistence installation of Porteus by `--ext4-installation` should look like.

[M. Eerie]

The problem is that I used a damaged pendrive and the creation of the second partition failed. I modified the script so that the first partition is of type 1b and 32 Mb because I like to have an older version of the kernel at hand.

For the time being I'm discarding the “net” version because if you look closer at my log file, I had downloaded an ISO whose checksum was “KO”, so the script downloaded the next one which was “OK”. Maybe that's why the checksum's deletion across the mirrors. Disturbing...

[robang74]

[...] if you look closer at my log file, I had downloaded an ISO whose checksum was “KO”, so the script downloaded the next one which was “OK”. Maybe that's why the checksum's deletion across the mirrors. Disturbing...

Someone might have tried my kmap-change script that alter the ISO using a mere `sed` hence on userland, on the server on which permissions were not set correctly...

• https://github.com/robang74/porteus-usb ... pchange.sh

For the time being I'm discarding the “net” version because [...]

I agree that usb-install script is much MORE reliable and safer to use at this stage of development. Which is the reason because net-install need to be tested.

Minghia...

In Italian is written with C not G and it means d1ck, but it means a neutral "wow". Where "neutral" means that can be used for a "WTF" (negative) or "minchia che bello!" (positive).

In which sense did you cast it at the end of your comment? Because you discovered a ISO which was sha256sum KO or because you appreciated my script!?

[M. Eerie]

In Italian is written with C not G

You are wrong.



I once worked with an Italian friend of mine from Nokia. He taught me that the origin was Sicilian and that it was written both ways. The literal translation in Spagnolo would be: "¡Cojones!"

In Spanish “del verbo” Spanish (...), the sense I wanted to give it is that of surprise/hastiness. Maybe the same one that Tom Cruise liked so much when he discovered the way to say it in my language: ¡Joder!”

"Todos los días, son días de aprender". Even italian

[robang74]

You are wrong.

I am not sure about that...

• https://en.wiktionary.org/wiki/Category:Sicilian_lemmas

Technically speaking Sicilian is a dialect and I was referring to italian

In Italian is written with C not G

because I was not sure that it was the same word, but yes if it comes from Sicilian dialect, it is the same word.

Curiously, English people who were working with me in Germany were also interested in bad words. It seems that Anglosakson are quite amused by the creativity of Southern Europe people in using language with "coloring" words. In this your experience match with mine. So, why you were surprisingly worried?

[robang74]

NET INSTALL V0.3.2 FOR PORTEUS v5.1 ALPHA

Release v0.3.2 is committed and ready to be tested by expert users and developers. It is based on v5.1 alpha, because v5.0.1 does not anymore SHA256sum.

• howto: https://github.com/robang74/porteus-usb ... ick--dirty

The easiest test to try is with the `net-install` on-demand by remote:

Code: Select all

mkdir usbtest; cd usbtest
bash <(wget -qO- tinyurl.com/27om5cm3)

Just few more information:

• the LIVE on VFAT seem faster in boot because the EXT4 install leverage lazy options on mkfs.ext4.

• performances greatly varies among USB sticks, some are more for VFAT and others better on EXT4 journaled

• The Porteus v5.1 alpha is slower in boot than the v5.0.1 and it easy to catch the difference

In the next version I will try different ways to deal with filesystems, to find a more suitable set of options.

However, 0.3.2 is going to stay because I have another project to carry on.

Therefore, I will care about developing further the v0.3.2 only in case of bug fixing for a little while in the future.

[robang74]

the sense I wanted to give it is that of surprise/hastiness

I suggest Porteus DEV team to sign EVERY files with a SHA256 sum, not just the ISO. Moreover creating a file with all the signatures would be very useful. Because, it is important that everyone can check the SHA256 sum of all the components before installing it and even after having installed it. Just in case. Moreover, that file SHOULD be shared also in the official Porteus website and signed with SHA256 sum itself. If you provide a github account for signature, even better.

How to deal with the version? Well, the file above will simply keep in consideration the last version available for a specific Porteus release - for example v5.0.1 - and each release will have one like Porteuse-v5.0.1-all-components-sha256sum.txt - for each version - Unfortunately, this is not sufficient because a compromised system can avoid to leverage the signature to keep itself secure (homologation). Hence, for each version of each components, it is needed to keep a SHA256 sum signature.

This is the reason because a github account is the best choice. Many people can fork, and git will grant the history plus github will grant timestamps. The github should imitate the standard mirror directory structure. With all the names of the components. For example:

• https://mirrors.dotsrc.org/porteus/x86_ ... 2-xtra.xzm

In this case the "https://mirrors.dotsrc.org/porteus" is the url header. Everything else is the path. So, in the github accout there will be, for example:

• mirrors/x86_64/Porteus-v5.0/modules/002-xtra.xzm.sha256sum

In this fashion, the $url_header will be replaced with the $rawgit_urlpath ending with "mirrors". This is fine for the last version of that component but many can be available. Fortunately, those available as updates are in a different path. Hence, the problem is solved by itself. On top of that, I strongly suggest to the Porteus DEV team to share also the main components into the /boot /EFI folders. In particular:

• dos/win executables and their dependencies, if any - into a .zip archive (or any other compression format)

• kernel/initramfs images separately, AFAIK because the kernel is currently separated

• boot and EFI folders, cleaned from the above components in a single compressed file

• the MBR which is the first 2048 512-blocks of the ISO image, but which as many block zeroed as possible (without the components above)

At this point creating an ISO from components became straight forward, and even better those components can be downloaded by a net-install application that will not even pass by a ISO image but creating a USB bootable image, or writing on the real device whatever it would be a USB stick or a SSD internal drive.

• the MBR for an installed systems, but which as many block zeroed as possible (without the components above)

Please note, that the MBR which is needed for an installation (MBR 1MB, p1: EFIBOOT VFAT 16MB, p2: EXT4/XFS system and data) is different from the ISO's one (e.g. isolinux vs syslinux). Which implies that also that component should be created and shared. In order to avoid as less frequent updates, that component should be "clean" by dependencies. Like this:

• https://github.com/robang74/porteus-usb ... ble.mbr.gz

Because it is save on a github account, it does not need a SHA256SUM because the commit has (in full) works in such a way.

• commit 0123419a17948f685bb6809ca68f58aa5882f1a8

This is true because the file is just 1650 bytes long. Which make sense to download it again to check it with the installed one. For components that are way bigger, the SHA256 would be necessary to speed up the checking procedure.

I hope this helps, R-

[M. Eerie]

So, why you were surprisingly worried?

Because it's the second pendrive that fails me in a short time.

[robang74]

Because it's the second pendrive that fails me in a short time.

My usb-install script put a lot of pressure on that devices, and on some USB stick kernel show very crappy dmesg outputs.

• https://github.com/robang74/porteus-usb ... ed-choices

Which is the reason because I felt the duty to suggest a pendrive suitable for this role.

Anyway, I am going to change the usb-install script in such a way that will leverage sequential writes, mainly.

SUGGESTION: when using old, cheap and slow 2.0 USB stick, go for VFAT LIVE installation.

[ncmprhnsbl]

while not addressing every point raised in this (and the other topic) individually, i'll just give a little update on what's been done server side so far:
for v5.01:

• the base modules(001,002,003) located in <arch>/Porteus-v5.01/modules are now links to the latest version in <arch>/Porteus-v5.01/updates/core/<date> folder.

• have added <arch>/Porteus-v5.01/install-skeleton.tar.xz , which is just the iso sans kernel and modules

• fixing the missing sha256sums is pending

for v5.1apha2 (new dev release)

• have added x86_64/Porteus-v5.1/install-skeleton.tar.xz

[robang74]

i'll just give a little update on what's been done server side

Thanks, I will leverages these changes as soon as the sha256sum.txt will be restored.