Easy 4.2.1 interim-release

[BarryK]

I have uploaded EasyOS version 4.2.1, but only the "delta" file, which is just 338KB.

I wanted to test updating from 4.2 using the new small delta-file system, and it works great. You do need the latest scripts:

/usr/sbin/mount-img
/usr/local/easy_version/easy-update

...attached, with false ".gz" appended.

If you have Easy 4.2 installed, you can update. It sure is a pleasure the 338KB download, compared with the full 'easy-4.2.1-amd64.img' at 825MB!

Probably work also with earlier installed versions, though they might be setup with 'vmlinuz' and 'initrd' not in the working-partition. Whatever boot-manager is used now has to have the "wkg_uuid" (or wkg_dev or wkg_label) and "wkg_dir" kernel parameters. In other words, there might be some pain involved with updating.

Also, updating earlier versions, the update will download the full-size image file, as only have a 4.2-to-4.2.1 difference-file online. So might as well wait for 4.2.2.

Even with 4.2, the if you have /mnt/wkg/easy-4.2-amd64.img, it might not have the correct checksum. The 'mount-img' script changed the file checksum. The latest mount-img doesn't.

Have a few things to do, expect 4.2.2 as the official next release, in a day or two.

I have rewritten the updating instructions:

https://easyos.org/user/how-to-update-easyos.html

...might require a browser page refresh.

[williwaw]

@BarryK

ez update is unresponsive.

Screenshot(1).png (23.5 KiB) Viewed 1085 times

steps taken

easy 4.2 installed to USB with DD. Limine config working well

installed the two scripts posted above, overwriting original

save on shutdown........ rebooted

copied easy-4.2-amd64.imgPRISTINE to /mnt/sdc2/easyos and renamed to easy-4.2-amd64.img no go at that location, nor when placed at /mnt/sdc2/

Even with 4.2, the if you have /mnt/wkg/easy-4.2-amd64.img, it might not have the correct checksum. The 'mount-img' script changed the file checksum. The latest mount-img doesn't.

checksum should match https://distro.ibiblio.org/easyos/amd64 ... md5sum.txt ?

[TerryH]

Downloaded the easy-update & mount-img scripts, then updated using the easy-update. Update process went without issue. Now running EasyOS 4.2.1. Fantastic. Once again, thank you.

[BarryK]

@BarryK

ez update is unresponsive.
Screenshot(1).png

steps taken

easy 4.2 installed to USB with DD. Limine config working well

installed the two scripts posted above, overwriting original

save on shutdown........ rebooted

copied easy-4.2-amd64.imgPRISTINE to /mnt/sdc2/easyos and renamed to easy-4.2-amd64.img no go at that location, nor when placed at /mnt/sdc2/

Even with 4.2, the if you have /mnt/wkg/easy-4.2-amd64.img, it might not have the correct checksum. The 'mount-img' script changed the file checksum. The latest mount-img doesn't.

checksum should match https://distro.ibiblio.org/easyos/amd64 ... md5sum.txt ?

You would get that empty window if /usr/local/easy_version/easy-update is not executable.
You have to right-click on the script, choose "Properties..." then tick the "Exec" checkboxes.
Same for /usr/sbin/mount-img

Yes, the checksum should match that online md5sum.txt

[BarryK]

Wanting to make the update more "bullet proof", I have put in an extra check in /usr/local/easy_version/easy-update
It now checks the md5sum of the local pre-update image file.

That is, if you are currently running 4.2 and have /mnt/wkg/easy-4.2-amd64.img, the 'md5sum.txt' file will be downloaded and the image file checked.

It has to be correct, otherwise the difference-file won't work.

After an update to, say, 4.2.1, you will then have /mnt/wkg/easy-4.2.1-amd64.img and also /mnt/wkg/md5sum.txt
They can both be left there in anticipation of the next update.

Latest easy-update attached, with false ".gz" appended...

[BarryK]

Yes, Limine is great.

Anyone interested in exploring the possibilities with the menu in 'limine.cfg'?

Starting with the usb-stick, menu entries could be added to boot Windows or any other installed Linuxes.

Limine is light-on in regard to examples.

The aim could be to develop "LimineConfig", a GUI bootloader installer, as a competitor to the Grub installers that have been developed for Puppy.

There is of course the venerable grub4dos installer:

https://distro.ibiblio.org/easyos/noarc ... -easy3.pet

Here is shinobar's old posts on grub4dosconfig:

https://oldforum.puppylinux.com/viewtopic.php?t=51697

Recently, shinobar has developed a grub2 installer PET, grub2config:

viewtopic.php?t=3360

But a first step would be to just play with 'limine.cfg' on the usb-stick, see if it can be made to recognize whatever else is installed on the PC.

[williwaw]

downloaded newest easy-upate.gz and gave it execute privileges this time :oops:

it ran fine until

hot.png (49.56 KiB) Viewed 973 times

Went to the working folder and inspected the new easy.sfs and it looked real close to the same size as the old...
so went with it to see if it would boot in spite of the error.

A boot of the fresh system brought up the same password request as before.....does the new initrd get updated with the old password?
But I never got past the error and failure to boot that occurred after the password was entered.*

so I manually replaced intird and vmlimz from the easy-updated/reconstructed easy-4.2.1-amd64.img, leaving the "presumably corrupt" easy.sfs in place, and deleted the other folders in the working partition for fresh frugal install of 4.2.1

new initrd offers options without the password enabled
snapshot works on the temp file without an extra reboot, if a manual save is made before the snapshot is taken
rollback works fine if elected at initrd runlevel.

rollback also works fine if added to a stanza on the kernel line in limine.cfg (It always has)

Code: Select all

    KERNEL_CMDLINE=initrd=initrd rw qfix=bak wkg_uuid=376xxxxxxxxxxxxxxx wkg_dir=easyos/

Code: Select all

QUIET=no
TIMEOUT=no

* lenovo T420 had some occasional failures to boot before updating, but became three failures in a row after update
will report later if problem resurfaces.

[BarryK]

williwaw,
Is that a usb-stick you are booting on?

"This file did not copy correctly: easy.sfs"

The update script copies easy.sfs out of the downloaded easy-4.2.1-amd64.img file, then runs an md5sum check comparing original with the copy.

That is done as a precaution, in case you have failing media.
Which looks to be your situation.

You can test for yourself. Open up the .img file and run md5sum on easy.sfs, then do same on /mnt/wkg/easy.sfs

The password is not stored in the initrd, it is not stored anywhere.
The folders /mnt/wkg/releases, etc. are encrypted by the password. When you enter the correct password, they become unencrypted and can be used.

The init script in the initrd does a test on one of those folders. If encrypted, the folder name, example 'releases' will be garbage text, so the folder doesn't exist. So the init script knows that a password has previously been applied to it, and then asks for the password.

[BarryK]

williwaw,
Another possibility is insufficient room in the working-partition. How much free space does it have?

[williwaw]

Barry,

the recent testing has all been done on a well used USB stick, a san disk ultrafit. the low profile ones that get rather hot.

the boot failures I have experienced may well also be on account of the USB wearing out. Normally, I tether a sata laptop sized ssd, because of the better controllers, and have a new one on hand which will be set up with Limine for a boot loader and a fresh install of 4.2.2

downloading now.........

[BarryK]

Barry,

the recent testing has all been done on a well used USB stick, a san disk ultrafit. the low profile ones that get rather hot.

the boot failures I have experienced may well also be on account of the USB wearing out. Normally, I tether a sata laptop sized ssd, because of the better controllers, and have a new one on hand which will be set up with Limine for a boot loader and a fresh install of 4.2.2

downloading now.........

I have one of those tiny usb-drives, but don't use it. I was alarmed how hot it got, and I also read that it can damage the usb-socket -- that might have just been someone's conjecture, don't recall.