Debian bookworm (12) image 1.0 released for B3

[MouettE]

Hello everyone,

I've just released the debian bookworm image for the B3. This time I tried to anticipate and be able to release it shorlty after the official release. As usual for the debian images, this image is constructed from 3 different projects : the bookworm OS image itself, an install/rescue system and an installer script. Also I've done a lot of work on the kernel package which had trouble booting during my first tries. More details on these further in this post.

Current users of the bullseye image can also directly upgrade their systems using standard system tools. More details below.

First the link and quick howtos for install and upgrade:

• b3-bookworm-install-1.0.zip (sha1: 5fbd744531d4785da95f01bc8cc7fe166aeab60d)

Quick install howto

This works the same way as the bullseye image:

• Unzip the downloaded file on a fat-formatted usb key with a partition table
• By default the installer will wipe the entire disk. If you wish to keep the existing data of your device, edit the install.ini file inside the install directory and set wipe to false at the bottom of the file. The installer will then only format the first partition of the drive.
• Plug the usb key on the B3 and apply power while pushing the rear button.
• The rescue/install system will start, format the drive, extract the bookworm image and reboot the server.
• By default the network is configured for dhcp on both network interfaces. It can be changed in the install.ini file. The installer will copy the network settings from install.ini to the bookworm system by default. This can also be changed in the install.ini file.
• If the LED turns red, something went wrong. Turn off the server and post the install.log file which is created in the install directory of the key so I can help you out.
• When the LED is solid blue you can ssh into the box with user 'excito' (no quotes) and password 'excito' (no quotes). Then you can 'su' ; the root password is 'excito' (no quotes). Note that on the first boot the system generates new ssh keys, it can take a minute or two before the ssh server is reachable after the LED turns blue.
• Once you're logged in I strongly recommend to set the locale and timezone of your server (its defaults are POSIX and UTC). Run the following commands :
  

  Code: Select all

  dpkg-reconfigure locales
  dpkg-reconfigure tzdata
  

Quick upgrade howto

I recommend you go through the bookworm release notes which include upgrade instructions. The only b3 specific task you need to do is to update the /etc/apt/sources.list.d/excito.list, change bullseye to bookworm inside when preparing apt sources (step 4.3 of the upgrade guide) and manually upgrade the kernel. If you use non-free firmware (who knows), you need to add the non-free-firmware component in your sources.list.

That being said if you want to go fast and don't mind fixing broken stuff after the upgrade here is a very quick-and-dirty how-to :

• update packages to the latest version with apt update and apt upgrade
• replace bullseye with bookworm in /etc/apt/sources.list and /etc/apt/sources.list.d/excito.list
• Replace the bullseye security repository definition with

  Code: Select all

  deb https://deb.debian.org/debian-security bookworm-security main contrib

• run

  Code: Select all

  apt update

  and

  Code: Select all

  apt full-upgrade

• Due to changes in kernel package numbering, run the following command to forcibly upgrade the kernel:

  Code: Select all

  apt install -y --allow-downgrades bubba3-kernel=`apt-cache show bubba3-kernel | grep "Version: 6" | cut -f2 -d' '`

• reboot the b3, check the running kernel with uname -a (should be a 6.1.x kernel) and check services with systemctl status
• Fix whatever needs to be fixed
• run

  Code: Select all

  apt autoremove --purge

  to cleanup old packages.

OS image contents

The bookworm OS is a very minimal debian install with only the core packages needed to start and ssh into the server. The image was bootstraped directly from standard debian utils (not built from upgrade of a previous version). There is a github repository which describes all the steps taken to build it.

Linux kernel

The kernel used is Linux LTS 6.1.x series with Excito patches. This is actually the series used in debian and it's LTS upstream. Sources are available on github. I spent quite some time to make the kernel run on the B3, mainly due to the old UBoot version. I tried to run the official debian kirkwood kernel from an interstitial kernel but tests were not successful. That's why I'm keeping the classic patched package.

Here are the most notable changes in the kernel package :

• Patches upgrading the dtc library (this one and this one) have been reverted because they are apparently not compatible with the U-Boot version (both 2010 and 2013 versions). Those reverts do not prevent the kernel from booting but it may complicate things on the next major upgrade.
• I've created a special minimal configuration which produces (I think) the smallest possible bootable kernel on the b3 (~780ko). Obviously it's not capable of booting any OS but allows to check that U-Boot is capable of booting the kernel version tested. As I had to build many many times looking for the problematic commits it saved me quite a lot of time. I've kept it in the source tree but it's obviously not used for the official build.
• I've completely reviewed the configuration and added many modules which could potentially be useful (USB mostly) This has increased the build time a lot on the B3 (22h now, it was 8h for the 5.10 series) but it's more complete now. I initially tried taking the debian configuration for kirkwood but the build time was craaazy long and most of the stuff was useless on the B3 so I simply added stuff in the old configuration. Nevertheless if you miss anything ping me and I'll add it to the configuration.
• I took some time rewriting large parts of the debian/rules build script. It was initialy a stripped down vesion of the official debian package (way up to the squeeze version) so I upgraded it based on the current bookworm official version (that also fixed parallel build which was broken on bookworm). The main goal was to fix the kernel headers package building which never worked (at least AFAIK), the challenge being to be able to cross-build them. Fortunately the official package allows it so I took the needed portions and added them to the bubba3 package. Now the headers are properly built both natively and crossly ! That way you can compile a custom module directly on the B3 (it was on my todo list from quite some time, it will probably never be used but it's satisfying and was a lot of fun).
• I ran litian on the package and tried to fix as much as I could.
• I've setup my b3 at home to automatically build and quickly validate the kernel when there is a new release in the 6.1 series. What it does is basically do a build, install the package locally and reboot. If the reboot is successful and the wired network is working, the package is pushed to the repository. That way it will be kept up with upstream (esp. for security patches) on a regular basis. If you think this approach is harsh, feel free to comment it here, I'm open to discussion for alternatives !

Miscellaneous

• Excito-specific packages (kernel and button and led management) are available in a debian repository I maintain on the B2BCS hosting servers. The image is pre-configured to use it so further upgrades can be installed with apt tools. Everything is on github too.
• U-Boot tools are installed and configured so you can play with u-boot if you dare. WARNING this is risky business, use it at your own risk.
• The root password is 'excito' (without quotes) and I've also created an excito user with the same password. Note that by default you can not ssh directly as root with a password.
• SSH host keys are generated on the first boot and it takes a few seconds.
• The default network configuration is DHCP for both interfaces. It may be overridden by the install.ini configuration.
• Due to the shutdown of the easyfind service, the easyfind-client package has not been ported to this version.

The install/rescue system

Full sources and rudimentary compilation instructions are available on github.

The installer script

The installer script has been ported to python 3.x following removal of version 2.x in buildroot. One bug has also been fixed. The rest is pretty much the same as before and is still configured through the install.ini file on the USB install key. It's heavily commented so all the options can be understood and changed accordingly.

Also in rescue-only or if reboot is disabled, the install script on exit will also still animate the LED to dictate the first IP address of the box. It's bonus and will hopefully make it easier to know the DHCP-attributed address of the server.

Conclusion

As always ask as many questions as you want I will be more than happy to answer them.

Charles

[elsbernd]

Hello MouettE,
I just did it ... I destroyed my original Excito-Installation after more than ten years. Since I new, that bullseye could be installed on my backup-B3 I tried this with my main B3, but used the wrong image (buster) two days ago fully configured for my usage (mainly easyfind, timeachine and ftp server). After I realised, that it was not bullseye, I tried a second time, configuring all again.
Only to realize, that you released bookworm. Becaus I'm a bookworm, the third installation followed.
Only some remarks:

• I don't know, how long I have to push the rear button, when powering up the system. I saw, that first the led went green and the ethernet speed leds and some internal leds in red were blinking. Then the blinking of all leds stopped, and it seemed, that the system was restarting.
  I still pressed the button. Then blinking began again and after more than 10 seconds I released the button.
  Is this too long?
• Since the installation went smooth, I had to install a small mariadb. Which failed. I assume, not engough memory.
• Couldn't find easyfind for myownb3.com any more, but I read, that this has been abonndoned. This was one of the main arguments running b3 for such a long time, but time goes by. I'll look at the thread.
• When installing bookworm, I failed to edit install.ini file. I added the parameter "size =", and forgot to add "full". My fault. But installation failed with:

  Code: Select all

  2023-06-13 06:46:26.852 - INFO - Running 'sgdisk -i 2 /dev/sdb'
  2023-06-13 06:46:26.890 - ERROR - Malformed float value for size in section general
  2023-06-13 06:46:26.892 - ERROR - Exception in the main install function :
  Traceback (most recent call last):
    File "/opt/excito/utils.py", line 286, in getfloat_check_conf
      res = config.getfloat(section, option)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/usr/lib/python3.11/configparser.py", line 839, in getfloat
    File "/usr/lib/python3.11/configparser.py", line 824, in _get_conv
    File "/usr/lib/python3.11/configparser.py", line 819, in _get
  ValueError: could not convert string to float: ''
  
  During handling of the above exception, another exception occurred: 
  
  Traceback (most recent call last):
    File "/opt/excito/installer", line 249, in <module>
      do_install()
    File "/opt/excito/installer", line 144, in do_install
      size = utils.getfloat_check_conf('general', 'size', 10.)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/opt/excito/utils.py", line 301, in getfloat_check_conf
      raise InvalidConf('Malformed float value for %s in section %s' % (option, section))
  utils.InvalidConf: Malformed float value for size in section general

It's my fault, so don't bother.

I'm glad, that you supported the B3 for all those years.
Klaus

[Gordon]

Although there is a power symbol next to it, it's actually not a power button. It requires an app to detect its state and do something in response ¹. Thus, once the LED turns green you can hold it for as long as you want but it won't do anything until the system is fully booted and bubba-buttond daemon is started.

I don't think there are additional red LEDs on the board? That's probably the hard drive. If the unit automatically reboots while in service mode it means it doesn't recognize the USB.


¹) interestingly the routine that handles the button's state in u-boot is a lot more CPU intensive than the user space daemon. Thus if you 'power down' the B3 and leave it connected to the PSU it can run pretty hot.

[MouettE]

I don't know, how long I have to push the rear button, when powering up the system.

Very shortly. Pusing the button simply alters the booting source when the b3 powers on (or reboots). The LED either starts green (button pushed, USB boot) or purple (released button, disk boot). As soon as you seen a green LED, you can release the power button.

Since the installation went smooth, I had to install a small mariadb. Which failed. I assume, not engough memory.

That is bound to happen eventually. You need to tune things up, you can refer to this. Let us know if and how you manage to run it.

Couldn't find easyfind for myownb3.com any more, but I read, that this has been abonndoned. This was one of the main arguments running b3 for such a long time, but time goes by. I'll look at the thread.

There are many alternatives even free ones. Don't hesitate to ask on the easyfind thread we can discuss it there.

When installing bookworm, I failed to edit install.ini file. I added the parameter "size =", and forgot to add "full". My fault. But installation failed with:

Well that is not a pretty way for the script to fail but nevertheless effective...

I'm glad, that you supported the B3 for all those years.

You're welcome !

[MouettE]

Although there is a power symbol next to it, it's actually not a power button. It requires an app to detect its state and do something in response

Technically you're right ; From an hardware perspective, the rear button is a GPIO-wired generic key. However in the B3 DTS it is marked as a power button :

Code: Select all

[...]
	gpio-keys {
		compatible = "gpio-keys";
		pinctrl-0 = <&pmx_button_power>;
		pinctrl-names = "default";

		power-button {
			/* On the back */
			label = "Power Button";
			linux,code = <KEY_POWER>;
			gpios = <&gpio1 7 GPIO_ACTIVE_LOW>;
		};
	};
[...]

Thus the OS considers it as a power button. You don't even need bubba-buttond (at least with systemd), the system will turn off immediately if the button is pushed. I wanted to keep the 2 seconds delay before powering off so I kept the bubba-buttond service but I needed to disable the systemd handling of the power button event (in the previous releases there was no delay as systemd bypassed bubba-buttond).

I don't think there are additional red LEDs on the board? That's probably the hard drive.

Yes it is, it's visible from the holes behind the b3.

[Gordon]

Although there is a power symbol next to it, it's actually not a power button. It requires an app to detect its state and do something in response

Technically you're right ; From an hardware perspective, the rear button is a GPIO-wired generic key. However in the B3 DTS it is marked as a power button :

Code: Select all

[...]
	gpio-keys {
		compatible = "gpio-keys";
		pinctrl-0 = <&pmx_button_power>;
		pinctrl-names = "default";

		power-button {
			/* On the back */
			label = "Power Button";
			linux,code = <KEY_POWER>;
			gpios = <&gpio1 7 GPIO_ACTIVE_LOW>;
		};
	};
[...]

Thus the OS considers it as a power button. You don't even need bubba-buttond (at least with systemd), the system will turn off immediately if the button is pushed. I wanted to keep the 2 seconds delay before powering off so I kept the bubba-buttond service but I needed to disable the systemd handling of the power button event (in the previous releases there was no delay as systemd bypassed bubba-buttond).

Yes, I know. It's actually a pretty bad idea to let systemd shut the system down like that though because as said the system doesn't know how to actually power down and it effectively just crashes into a state where you need to disconnect and then reconnect the power cord.

[elsbernd]

Hello again.
Since I own 4 Excitos B3 I bought a 4TB SSD for testing.
Installation went smooth, but rebooting failed. Here is the intall.ini and protocol:

Code: Select all

[general]
image = bookworm/bookworm-image-1.0-20230611-1733.tar.xz
wipe = true
size = full
swap = 1024
reboot = true

Code: Select all

2023-06-19 04:27:24.316 - INFO - Running Excito installer version 2.2 daemon process with pid 95
2023-06-19 04:27:24.329 - INFO - Checking image file /mnt/usb/install/bookworm/bookworm-image-1.0-20230611-1733.tar.xz
2023-06-19 04:27:24.337 - INFO - Checking /dev/sdb
2023-06-19 04:27:24.338 - INFO - Running 'fdisk -l /dev/sdb'
2023-06-19 04:27:24.404 - INFO - Running 'sgdisk -p /dev/sdb'
2023-06-19 04:27:24.477 - INFO - Running 'blkid /dev/sdb1'
2023-06-19 04:27:24.552 - INFO - Running 'blkid /dev/sdb2'
2023-06-19 04:27:24.622 - INFO - Running 'sgdisk -i 1 /dev/sdb'
2023-06-19 04:27:24.659 - INFO - Running 'sgdisk -i 2 /dev/sdb'
2023-06-19 04:27:24.697 - INFO - Destination disk: sdb (3.6 TiB, gpt)
2023-06-19 04:27:24.699 - INFO - Wiping existing partition label on sdb
2023-06-19 04:27:24.701 - INFO - Running 'sgdisk -Z /dev/sdb'
2023-06-19 04:27:24.734 - ERROR - [sgdisk] Warning! Unable to generate a proper UUID! Creating an improper one as a last
2023-06-19 04:27:24.736 - ERROR - [sgdisk] resort! Windows 7 may crash if you save this partition table!
2023-06-19 04:27:25.779 - INFO - [sgdisk] GPT data structures destroyed! You may now partition the disk using fdisk or
2023-06-19 04:27:25.780 - INFO - [sgdisk] other utilities.
2023-06-19 04:27:25.785 - INFO - Creating GPT label, system 3725.0 GiB size partition and swap 1024.0 MiB size on sdb
2023-06-19 04:27:25.786 - INFO - Running 'sfdisk /dev/sdb' with this input: label: gpt|,3725.0GiB,L|,,S|
2023-06-19 04:27:25.890 - INFO - [sfdisk] Checking that no-one is using this disk right now ... OK
2023-06-19 04:27:25.892 - INFO - [sfdisk] 
2023-06-19 04:27:25.893 - INFO - [sfdisk] Disk /dev/sdb: 3.64 TiB, 4000787030016 bytes, 7814037168 sectors
2023-06-19 04:27:25.894 - INFO - [sfdisk] Disk model: Samsung SSD 870
2023-06-19 04:27:25.895 - INFO - [sfdisk] Units: sectors of 1 * 512 = 512 bytes
2023-06-19 04:27:25.896 - INFO - [sfdisk] Sector size (logical/physical): 512 bytes / 512 bytes
2023-06-19 04:27:25.897 - INFO - [sfdisk] I/O size (minimum/optimal): 512 bytes / 512 bytes
2023-06-19 04:27:25.898 - INFO - [sfdisk] 
2023-06-19 04:27:25.899 - INFO - [sfdisk] >>> Script header accepted.
2023-06-19 04:27:25.900 - INFO - [sfdisk] >>> Created a new GPT disklabel (GUID: 987579D9-1A64-BD47-9B5D-5A94B5F69A8C).
2023-06-19 04:27:25.906 - INFO - [sfdisk] /dev/sdb1: Created a new partition 1 of type 'Linux filesystem' and of size 3.6 TiB.
2023-06-19 04:27:25.905 - ERROR - [sfdisk] Partition #1 contains a ext3 signature.
2023-06-19 04:27:25.931 - ERROR - [sfdisk] Partition #2 contains a swap signature.
2023-06-19 04:27:25.932 - INFO - [sfdisk] /dev/sdb2: Created a new partition 2 of type 'Linux swap' and of size 1 GiB.
2023-06-19 04:27:26.653 - INFO - [sfdisk] /dev/sdb3: Done.
2023-06-19 04:27:26.654 - INFO - [sfdisk] 
2023-06-19 04:27:26.655 - INFO - [sfdisk] New situation:
2023-06-19 04:27:26.656 - INFO - [sfdisk] Disklabel type: gpt
2023-06-19 04:27:26.657 - INFO - [sfdisk] Disk identifier: 987579D9-1A64-BD47-9B5D-5A94B5F69A8C
2023-06-19 04:27:26.658 - INFO - [sfdisk] 
2023-06-19 04:27:26.659 - INFO - [sfdisk] Device          Start        End    Sectors  Size Type
2023-06-19 04:27:26.660 - INFO - [sfdisk] /dev/sdb1        2048 7811893247 7811891200  3.6T Linux filesystem
2023-06-19 04:27:26.663 - INFO - [sfdisk] /dev/sdb2  7811893248 7814035455    2142208    1G Linux swap
2023-06-19 04:27:26.664 - INFO - [sfdisk] 
2023-06-19 04:27:26.665 - INFO - [sfdisk] The partition table has been altered.
2023-06-19 04:27:26.666 - INFO - [sfdisk] Calling ioctl() to re-read partition table.
2023-06-19 04:27:26.667 - INFO - [sfdisk] Syncing disks.
2023-06-19 04:27:26.670 - INFO - Disk looking good, proceed with formatting
2023-06-19 04:27:26.671 - INFO - Formatting system partition
2023-06-19 04:27:26.672 - INFO - Running 'mkfs.ext3 -Fq /dev/sdb1'
2023-06-19 04:32:45.172 - INFO - Formatting swap partition
2023-06-19 04:32:45.173 - INFO - Running 'mkswap -f /dev/sdb2'
2023-06-19 04:32:45.188 - INFO - [mkswap] mkswap: /dev/sdb2: warning: wiping old swap signature.
2023-06-19 04:32:45.211 - INFO - [mkswap] Setting up swapspace version 1, size = 1 GiB (1096806400 bytes)
2023-06-19 04:32:45.212 - INFO - [mkswap] no label, UUID=55af99ab-655f-418d-870b-b56086055a5b
2023-06-19 04:32:45.214 - INFO - Mounting system partition
2023-06-19 04:32:45.215 - INFO - Running 'mount /dev/sdb1 /mnt/target'
2023-06-19 04:32:45.635 - INFO - Extracting image file to target
2023-06-19 04:32:45.636 - INFO - Running 'tar -xf /mnt/usb/install/bookworm/bookworm-image-1.0-20230611-1733.tar.xz -C /mnt/target'
2023-06-19 04:34:13.914 - INFO - Copying resolv.conf to the target
2023-06-19 04:34:13.924 - INFO - Generating network settings to target with image-supplied template
2023-06-19 04:34:13.928 - INFO - Umounting system partition
2023-06-19 04:34:13.929 - INFO - Running 'umount /mnt/target'
2023-06-19 04:34:14.831 - INFO - System successfully installed
2023-06-19 04:34:14.832 - INFO - install done; rebooting system
2023-06-19 04:34:14.860 - ERROR - Error while running excito installer
Traceback (most recent call last):
  File "/opt/excito/installer", line 258, in <module>
    sys.exit(0)
SystemExit: 0

Any ideas, what happened.
powering off and on without USB-drive gives violet LED.

[MouettE]

Hello and thanks for the feedback.

The error at the end of the installer log is harmless, it shouldn't happen (new bug) and I will correct it later. From what I see in the logs, the install went very good. It's (very) likely that U-Boot is incapable of booting from such a large drive or partition. I don't know if the issue comes from the disk itself being that big or the partition. In the second case, reducing the size of the partition should suffice (let's say 20G which should be more that enough for the b3 ; feel free to go above but don't go over 2Tb). If it's the disk then I suggest you upgrade UBoot following the dedicated wiki page.

[Gordon]

Violet/purple LED means that as far as u-boot is concerned everything is okay. If it doesn't recognize the disk the LED will turn red, i.e. with the original u-boot it will, but I would seriously doubt that this could recognize a 4TB SSD. So I'm guessing that you already have the community u-boot and that pretty much means that all odds are off with respect to the LED reflecting a faulty disk state. I actually hacked one of my B3s to get around such an issue, making it boot from USB and then transfer control to the internal hard drive using kexec.

[elsbernd]

Thank you both Gordon/MouettE,
It seems, I'm getting old. Cound'nt imagine that there is a system out there with the ancient 2TB limit. But well, my first B3 is from 2012. I Just installed it with 1800GB.
Yes I know, 20GB is sufficient, but I'll try to create a system as a backup-System for my actual data on a server. I could directly attach the SSD to the server, but it's more fun to use it on the B3.
With 1800GB it's up and running. And he's fast enough for playing.
Rem: I had to disconnect ethernet and reconnect to convice the system to actually get an ip-address from dhcpd after the installation finished.

My B3s won't go into retirement. They'll get power until they die.
And well, I'm not that old. My actually servers (2012, 2018, 2022) are all running the actuall OpenIndiana

[Legobas]

Thank you so much for providing this image!
It makes me very happy to have an up-to-date Debian on my B3.

The install procedure went flawlessly, I choose to create a freshly new Debian (wipe = true)

[nmichen]

Massive thanks for keeping my pool of B3s up and running latest Debian!!

I ran the update track from the previous version and it all went smoothly except for the kernel package where apt believes 1:5.10.xyz is newer than 6.1, so has to force a "downgrade" to 6.1.

Anyway, I run my services (samba, nfs, home made services, ..) in docker containers to reduce dependency on the host versions. Unfortunately docker depends on the devicemapper thin pool feature which I always compiled my self, but now that the kernel is updated frequently this becomes a bit cumbersome. Could you please add that to the default kernel build?

Again, massive thanks for the support. My 4 (launch time) B3s has served me very well over the years and continue to be my low power multi-offsite backup solution. I'm an electronic engineer by trade and have played with the idea of producing a new PCB with newer CPU that could be a slide in replacement and provide it open source so anyone cold order a new from any online PCB manufacturers. Just have to get the time (limited with small children).

[MouettE]

Massive thanks for keeping my pool of B3s up and running latest Debian!!

Thank you, good to know it's useful !

I ran the update track from the previous version and it all went smoothly except for the kernel package where apt believes 1:5.10.xyz is newer than 6.1, so has to force a "downgrade" to 6.1.

OK, I will verify and update the first post, thanks for the feedback.

Anyway, I run my services (samba, nfs, home made services, ..) in docker containers to reduce dependency on the host versions. Unfortunately docker depends on the devicemapper thin pool feature which I always compiled my self, but now that the kernel is updated frequently this becomes a bit cumbersome. Could you please add that to the default kernel build?

I could definitely add it, can you confirm that it is the CONFIG_DM_THIN_PROVISIONING (Thin provisioning target) ? I'll start the build as soon as you'll confirm it.

I'm an electronic engineer by trade and have played with the idea of producing a new PCB with newer CPU that could be a slide in replacement and provide it open source so anyone cold order a new from any online PCB manufacturers. Just have to get the time (limited with small children).

That would probably interest a lot of people here including myself

[nmichen]

yes, CONFIG_DM_THIN_PROVISIONING is the one!

[Clive]

It's killed my B3, it won't reboot... Stays at the purple light, although lots of disk activity beforehand whilst attempting to boot.

Any ideas ? As much as I love my ancient B3, I am not sure how much energy I have to try and fix it....

It's pingable, but I am not able to ssh into it....