Home Home > Base-system
Sign up | Login

Deprecation notice: openSUSE Lizards user blog platform is deprecated, and will remain read only for the time being. Learn more...

Archive for the ‘Base System’ Category

Fun things to do with driver updates

February 16th, 2017 by

Today: update the update process!

Yesterday a colleague asked me if it would be possible to apply a driver update (DUD) to the rescue system. He wanted to use a new btrfsprogs package.

My immediate reaction was: no, you can’t do it. But then, there’s no technical reason why it shouldn’t be possible – it actually nearly works. The updates are downloaded as usual – just not applied to the rescue system.

So I thought: “Why not make a driver update so driver updates work also for the rescue system?”

Here’s how I did it.

First, let’s find out how driver updates are usually applied. The code is here:

https://github.com/openSUSE/installation-images/blob/master/data/root/etc/inst_setup#L84-L87

We need just these three lines:

for i in /update/[0-9]*/inst-sys ; do
  [ -d "$i" ] && adddir "$i" /
done

linuxrc downloads the driver updates and stores them in an /update directory. One (numbered) subdirectory for each update.

It obviously uses some adddir script. So we’ll need it as well. Luckily, it’s not far away:

https://github.com/openSUSE/installation-images/blob/master/data/root/etc/adddir

Next, we’ll have to find the spot where the rescue system is set up. It’s done in this script:

https://github.com/openSUSE/installation-images/blob/master/data/initrd/scripts/prepare_rescue

Let’s do some copy-and-paste programming and insert the above code near the end of the script. It then might look like this

# driver update: add files to rescue system
if [ -d /mounts/initrd/update ] ; then
  cp -r /mounts/initrd/update /
  for i in /update/[0-9]*/inst-sys ; do
    [ -d "$i" ] && /mounts/initrd/scripts/adddir "$i" /
  done
fi

Some notes:

  • You have to know that prepare_rescue is run as the last thing before we exec to init. So everything is already in place, the left-over files from initrd are mounted at /mounts/initrd and will be removed at the end of the script.
  • This means we have to copy our updates into the new root directory, else they will be lost.
  • Also, we plan to make the adddir script available at /scripts/adddir by our driver update (see below).

Now let’s create the driver update:

mkdud --create dud_for_rescue.dud \
  --dist tw --dist leap42.1 --dist leap42.2 --dist sle12 \
  --name 'Apply DUD also to rescue system' \
  --exec 'cp adddir prepare_rescue /scripts' \
  adddir prepare_rescue

Here’s what this call does, line-by-line:

  • the fix works for all current SUSE distributions, so let’s support them
  • give the driver update some nice name
  • this command is run right after the driver update got loaded; we copy the scripts out of the driver update to their final location
  • add adddir and our modified prepare_rescue script

Here is the result: dud_for_rescue.dud.

Now, back to the original problem: how to use this to update a package in the rescue system? That’s easy:

mkdud --create new_btrfs.dud \
  --dist sle12 \
  dud_for_rescue.dud btrfsprogs.rpm

creates a driver update (for SLE12) that updates btrfsprogs also in the rescue system.

How to build OS images without kiwi

December 26th, 2016 by

kiwi has long been the one standard way of building images in openSUSE, but even though there exist extensive writings on how to use it, for many it is still an arcane thing better left to the Great Magicians.

Thus, I started to use a simpler alternative image building method, named altimagebuild when I built our first working Raspberry Pi images in 2013 and now I re-used that to build x86_64 VM images at
https://build.opensuse.org/package/show/home:bmwiedemann/altimagebuild
after I found out that it even works in OBS, including publishing the result to our mirror infrastructure.
It is still in rpm format because of how it is produced, so you have to use unrpm to get to the image file.

This method uses 3 parts.

  • a .spec file that lists packages to be pulled into the image
  • a mkrootfs.sh that converts the build system into the future root filesystem you want
  • a mkimage.sh that converts the rootfs into a filesystem image

The good thing about it is that you do not need specialized external tools, because everything is hard-coded in the scripts.
And the bad thing about it is that everything is hard-coded in the scripts, so it is hard to share general improvements over a wider range of images.

In the current version, it builds cloud-enabled partitionless images (which is nice for VMs because you can just use resize2fs to get a larger filesystem and if you later want to access your VM’s data from outside, you simply use mount -o loop)
But it can build anything you want.

To make your own build, do osc checkout home:bmwiedemann/altimagebuild && cd $_ && osc build openSUSE_Leap_42.2

So what images would you want to build?

AMD/ATI Catalyst fglrx rpms, end of an era!

December 7th, 2016 by

Long time not talking about fglrx rpm, mostly because they’ve got no update since last December 2015.

Short Summary

In a word as hundred, fglrx is now a dead horse!

Dead horse

We had the hope of getting it working for Leap 42.2 in October, but except freezing kernel and xorg, you will not get what you would expect: a stable xorg session

Say goodbye fglrx!, repeat after me, goodbye fglrx.

If you are locked down and forced for any reasons to use fglrx with your gpu, and are still using 42.1, then don’t upgrade to 42.2, without a plan B

It has no more support from AMD upstream, and that’s it!, if someone want to break its computer, it’s still possible to pick the last files and try it by yourself, but the repository will never contain it for 42.2 (see below how-to)

That’s said, I’m not still sure, to keep for a long time the repository, I’ve been managing since 6 years now.

A bit of history

In 2010, when we were working hard to get 11.1 out, the news that no supported ATI (at that time) will be available for end-users, as we have for nvidia gpu

I didn’t check back the irc log, but we were a few, that would like to have this still available, by pure commodity. Especially that I’ve just exchanged a non working gpu by my new hd5750.

I remember the first chaotic steps, how to build that, and create repeating builds, what about the license? Did we have the right to offer a pre-build rpm etc. I spent some time fixing all of this stuff.
And start the build on real hardware. Hey afterward kvm was really in infancy stage.

Release after release amd/ati and openSUSE, the driver was build, on hardware for each supported distribution. When beginning of 2013 Sebastian Siebert, who got some direct contacts with AMD, release
his own script, we collaborate to have the possibility to build on virtual machines, which allow me to simplify the build process, as having on kvm for each openSUSE release supported.

Afterward, AMD start to split fglrx with the fglrx for HD5xx and above, and fglrx-legacy. So 2 drivers to maintain, but as always with proprietary software, the legacy version became rapidly obsolete,
and non usable. Not that bad, in the meantime the AMD effort on the free and open source radeon driver, quickly overcome the performance of legacy.

Still from 2013, to 2016 I’ve been able to propose ready to use rpm for several version of openSUSE’s distributions. I think the repository serve quite well end users, and I never got big flames.

I can’t avoid to mention the openSUSE powered server and sponsored by Ioda-Net Sàrl that has serve this objective so well during that time frame.

Future of the repository

Now that fglrx is becoming obsolete, I think seriously about why the repository online should stay online.

At openSUSE project level, we still have 13.1, 13.2, 42.1 and 42.2 that are mostly active. 13.1 is already almost out of the game of evergreen,
13.2 will follow soon, and I don’t know yet the exact plan for 42.1, but it will certainly go out of maintenance in less than a year.

If you feel or have the need of the repository, please express that in the comments below.

Wait there’s amd-gpu-pro, no?

Yeap there’s a closed driver, called amd-gpu-pro, available, for newer cards. But there’s two things that bring me out of the game, first I don’t have those newer gpu,
and don’t have the need to replace my hd5750 for the moment. The second and certainly the most important, those drivers are only available for Ubuntu or at least in .deb format.

I will certainly not help proprietary crap, if I don’t have a solid base to work with, and a bit of help from their side. I wish good luck to those who want to try those drivers,
I’ve got a look inside, and got a blame face.

For crazy, and those who don’t love their computer

So you want to loose your time? you can! I’ve kept in raw-src directory all the script used to build the driver.
They differ a bit compared to Sebastian Siebert last version in the sense of making Leap 422 as a possible target.
If you dig a bit around, you should be able to build them, but you’re alone on that way, you’ve been warned!

I’m not against a republished version, if someone find a way to make them working, just drop me a message.

That’s all for this journey, Have Fun! 🙂

Highlights of YaST development sprint 28

December 2nd, 2016 by

November is over, Santa Claus elves start to stress and the YaST team brings you one of the last reports of 2016. Let’s see what’s new in YaSTland.

Harder to ignore installation warning

The “installation settings” summary screen usually reports some non-critical errors displayed as a red text. Although the installation can proceed despite those errors, they are usually serious enough to lead to problems. That’s why we decided to introduce a change to highlight them a little bit more, making them harder to overlook.

The following screenshot shows the newly introduced confirmation dialog, presented before proceeding with installation.

Preventing users to shoot their feet

Make DHCLIENT_SET_HOSTNAME configurable on a per-interface basis

But that’s not the only usability-oriented enhancement on this sprint. We also reworked a bit the network configuration dialog.

For home users is very common to use a fixed hostname -set during installation- for our beloved linux box. But in some circumstances it’s better to set the hostname of the machine dynamically using DHCP, something YaST has always allowed to do by just ticking a checkbox that used to be in the network configuration screen. See “Change Hostname via DHCP” below.

The old network settings screen

That checkbox used to modify the system-wide variable DHCLIENT_SET_HOSTNAME, which was fine in scenarios in which only one of the network interfaces was configured via DHCP. But with several network interfaces connected to different DHCP-enabled networks, some problems arose.

During installation, if network configuration is used, Linuxrc creates the ifcfg files with DHCLIENT_SET_HOSTNAME='yes' for all of the enabled or configured interfaces and this value has precedence over the global one.

So the main problem was that YaST only allowed us to modify the global variable and setting it to ‘no’ did nothing because it was enabled for some interface.

During this sprint we have fixed that and now the user interface offers the possibility of choosing which DHCP interface will be used to decide the hostname.

The new network settings screen

Apart from choosing one of the existing interfaces, the new setting can also be set to ‘no’ or to ‘any’. In any case, YaST will always configure the system-wide options and the interface specific ones in a consistent way, so the behavior is always predictable.

But YaST is not the only way of configuring the network, so it’s always possible to have an unpredictable configuration. Fortunately, those potentially problematic scenarios will be detected by YaST and reported to the user.

Detecting dangerous scenarios in network settings

Partitioning in CASP

In the previous report we already explained how are we improving the installer to support the definition of the ultra-streamlined installation process of SUSE CASP, the new Kubernetes based member of the SUSE family.

In this sprint we introduced several additional changes to enable a different partitioning approach, more guided and automatic than ever. In a CASP node it makes no sense to use the advanced settings offered by our storage proposal, like encryption or LVM. Moreover, CASP relies on Btrfs to provide some of its cool and advanced features, like transactional updates.

As a result, although the regular SUSE and openSUSE releases will keep offering all the current possibilities in the same way than ever, in CASP the partitioning step will be skipped and the automatically calculated proposal will be simply displayed in the installation summary.

The new CASP installation summary

Clicking on the proposal will allow to re-target the installation to a different disk (or disks) in a similar way than the regular installer, but the options will be more limited. Again, no easy way to use LVM, encryption, separate home or any file system type other than Btrfs.

Selecting the partitions in CASP, no proposal settings button

The expert partitioner is still available during CASP installation, but using it will show an extra warning, since it implies a much bigger risk than using it in a regular SUSE or openSUSE system.

Expert partitioner warning in CASP

Improved debugger integration

We have improved the Ruby debugger integration in YaST. So far you could start the debugger using the y2debugger=1 boot option or by setting the Y2DEBUGGER=1 environment variable. The new feature allows starting the Ruby debugger also later when the YaST module is already running.

Simply press the Shift+Ctrl+Alt+D keyboard shortcut (D as debug) and it will start the Ruby debugger. It works during installation and also in installed system (just make sure the byebug Ruby gem is installed).

Unfortunately this new feature works only in the Qt UI, the ncurses UI is not supported (currently it does not handle the debugging keyboard shortcut at all).

After pressing the keyboard shortcut the debugger window will pop up:

New debugger integration

Storage reimplementation: it’s alive!

It took us one more sprint than originally expected, but finally we can say the testing ISO for the new storage stack is fully installable.

We fixed the UEFI + MBR partition table scenario we already had almost working in the previous sprint (turns out it was not that broken in Tumbleweed after all) and we adapted yast2-bootloader to be also able to deal with legacy (i.e. no UEFI) booting using the new storage stack.

As a nice result, our testing ISO can be used to install a perfectly functional system in both UEFI or legacy systems with the only requirement of having a pre-existing MBR partition table in the disk. It only shows a couple of error pop-ups related to the calculation of the proposal of software to be installed, but nothing that would prevent you from replacing whatever operating system you have with a new shiny openSUSE-based experiment.

This milestone opens the door to start testing the new stack with openQA, the same system that helps to guarantee the robustness of all the recent SUSE and openSUSE versions.

Storage reimplementation: preparations for the storage proposal

Now that yast2-bootloader starts to be ready to work with the new storage stack in more and more scenarios, it’s time to adapt the only component that still complains during the installation.

In order to make that task doable during the next sprint, we invested some time in this sprint analyzing the interaction between the software proposal calculator and the old storage layer. The outcome was a small document detailing what needs to be adapted in the proposal and in the new stack. The perfect input for a task in the next sprint.

Help for power-users with short memory

Our beloved YaST is packed with magic tricks below the surface. Many of them are very useful to debug installation problems or to better understand how the YaST internals work. Unfortunately developers tend to not be that good at blindly memorizing stuff and the functionality is so well hidden that most newcomers would have hard times finding it… until now.

We have added a couple of new keyboard shortcuts to show a summary of all the advanced hotkeys, so now you only have to remember one key combination instead of a dozen of them. In both text (ncurses) and graphical (Qt) mode, it will be enough to press shift+F1 to get the advanced help displayed below. Since some terminal emulators could already use that combination, ctrl+D F1 can also be used in the ncurses interface as an alternative.

Advanced Hotkeys help dialog

Contributions keep coming!

As we have already mentioned in previous sprint reports, an important part of our daily job as open source developers is helping casual (and not so casual) contributors to bring their ideas and code into YaST and related projects.

This time that (hopefully not casual) contributor was Devin Waas, who wanted to improve the installation to make the life of cloud-lovers easier.

For cloud guys out there retrieving logs of a failed installation is was a huge problem. Now, thanks to Devin, all you need is a running a rsyslog server and you’ll be able to easily access your installation logs from there.

A drawing is worth a thousand words

As a matter of fact, the newest Tumbleweed release allows us to specify the IP address of a remote server from the bootloader through the “Loghost flag”. Linuxrc will take care of setting up a UDP broadcast for dmesg contents and YaST installation logs.

This is just a first step. Devin promised further improvements of our newly implemented remote logging system. And he codes better than he draws, so stay tuned!

Storage reimplementation: LVM-based proposal

As we already mentioned in previous reports, when we started to develop the partitioning proposal we first focused in the scenario of a partition-based proposal with one or several MBR-style partition tables. That looked like the most complex scenario due to the limited number of primary partitions, the alignment problems, the overhead introduced by the EBR (extended boot record) of every logical partition and so on.

A couple of sprints ago, we got that working so we started to work on the LVM-based proposal. It took a little bit longer than expected but now we are able to generate LVM-based proposals for almost every possible scenario. The goal was to have them working in our mocked test cases. So probably the new LVM-based proposal cannot still be used to install a fully functional system, but it is backed by a full load of tests that prove we can handle many situations, from trivial to really tricky ones… and believe us, things can get quite tricky if you mix logical partitions with their EBR overhead and LVM volumes with their PE size rounding and their metadata overhead.

Bugs, bugs, bugs

In this sprint we kept the already commented approach of making the fix of low-priority and small bugs part of the Scrum process. As a result we accounted for approximately 50 deaths of those annoying creatures.

Conclusion

Looking at the report, we could say it was a quite successful sprint. But to be honest we were aiming even higher. Quite some interesting PBIs (features or bug-fixes in Scrum jargon) were almost done at the end of the sprint. But following Scrum philosophy, we never blog about almost-done stuff.

Thus, if nothing goes wrong things will be even better in the next report in three weeks. So have a lot of fun trying the new stuff and stay tuned for more!

Highlights of YaST development sprint 27

November 10th, 2016 by

Another three weeks of development come to an end… and our usual report starts. Take a look to what we have been cooking.

Read-only proposal modules

This week, during SUSECon 2016, SUSE announced an exciting upcoming new product. SUSE CASP – a Kubernetes based Container As a Service Platform.

That has, of course, some implications for the installer, like the need of some products (like CASP) to specify a fixed configuration for some subsystems. For example, an established selection of packages. The user should not be allowed to change those fixed configurations during installation.

We have implemented a possibility to mark some modules in the installation proposal as read-only. These read-only modules then cannot be started from the installer and therefore their configuration is kept at the default initial state.

Software and firewall proposals as read-only

In this sprint we have implemented a basic support in the proposal framework, in the future we could improve the respective proposal modules to better handle the read-only mode.

Per product Btrfs subvolumes handling

The ability to have read-only proposals is not the only improvement we have introduced in YaST to make SUSE CASP possible.

When Btrfs is used during the installation, YaST2 proposes a list of subvolumes to create. Unfortunatelly, that list is hard-coded and it’s the same for every (open)SUSE product… until now.

Starting on yast2-storage 3.1.103.1, a list of subvolumes can be defined in the product’s control file along with additional options (check out SLES and openSUSE examples to learn more).

<subvolumes config:type="list">
  <subvolume>
    <path>boot/grub2/i386-pc</path>
    <archs>i386,x86_64</archs>
  </subvolume>
  <subvolume>
    <path>home</path>
  </subvolume>
  <subvolume>
    <path>opt</path>
  </subvolume>
  <subvolume>
    <path>var/lib/libvirt/images</path>
    <copy_on_write config:type="boolean">false</copy_on_write>
  </subvolume>
</subvolumes>

This specification supports:

  • Disabling copy-on-write for a given subvolume (it’s enabled by default).
  • Enabling the creation of a subvolume only for a set of architectures.

Improved AutoYaST Btrfs subvolumes handling

AutoYaST Btrfs subvolumes handling has been also improved. Using almost the same syntax as for product control files, you can disable the CoW behavior for a given subvolume.

Of course, if you don’t need such a feature, you won’t need to adapt your profiles to the new syntax as it’s backward compatible. You can also mix both of them:

<subvolumes config:type="list">
  <subvolume>home</subvolume>
  <subvolume>
    <path>var/lib/libvirt/images</path>
    <copy_on_write config:type="boolean">false</copy_on_write>
  </subvolume>
</subvolumes>

On the other hand, if you’re running SLES, you’ll know that a subvolume called @ is used as the default Btrfs subvolume. Now is possible to turn-off such behavior in the profile’s general section.

<general>
  <btrfs_set_subvolume_default_name config:type="boolean">false</btrfs_set_subvolume_default_name>
</general>

Disable installer self-update by default

We have talked many times in this blog about the new self-update functionality for the installer. As expected, this functionality will debut in SLE-12-SP2 but with a small change in the original plan. In order to ensure maximum consistency in the behavior of SLE-12-SP1 and SP2 installers, the self-update functionality will be disabled by default. Not a big deal, since enabling it to get the latest fixes will be just a matter of adding a boot option in the initial installation screen.

Storage reimplementation: adapted EFI proposal in yast2-bootloader

The previous report about the status of the storage stack reimplementation finished with the following cliffhanger “we have in place all the ingredients to cook an installable ISO“. So, as expected, we worked during this sprint in adapting the bootloader proposal to the new storage layer. As you can see in the following screenshot, we succeeded and the installer can already propose a valid grub2 setup to boot an EFI system.

EFI proposal with the storage-ng ISO

Does that mean that the testing ISO for the new storage stack is already fully installable? Unfortunately not. Why not, you ask? The reason is, in fact, kind of fun.

The current partitioning proposal only works with MS-DOS style partition tables because we wanted to address the most complicated case first. On the other hand, we intentionally restricted the adaptation of the bootloader proposal to the EFI scenario. And you know what? We found out that combination (MS-DOS partition table + EFI) does not currently work in Tumbleweed, so it neither does in our Tumbleweed-based testing ISO.

We will work during the following sprint to support another scenario. And hopefully we will not choose again an unsupported or broken one. 😉

Letting libYUI run free: first visible steps

As you know, YaST has both a graphical and a textual interface. They run on the same code thanks to an abstraction layer called LibYUI. Originally it served YaST only, but over time other projects started using it, notably the admin panel in Mageia. Moreover, the developers of those projects started to contribute fixes and improvements to LibYUI… faster than we can cope with.

Recently we decided to give the people outside the YaST team more power in the LibYUI project. To make that possible ensuring it does not affect YaST stability, we have been enabling more continuous integration tests.

As a first fruit of the revamped collaboration, we have merged a fix contributed by Angelo Naselli of Mageia regarding selection of tree items. As soon as we complete our continuous integration setup (of course we will keep you informed), more fixes and improvements will come for sure.

More bug squashing and bug paleontology

In the previous report we presented our new effort to integrate the fix of low-priority bugs into the Scrum process. During this sprint we refined the idea a little bit more, distinguishing two separate concepts (each of them with its own PBI) – fixing of existing bugs and closing of too old ones.

The first one doesn’t need much explanation. We managed to fix around 25 non-critical bugs in YaST and if you are using Tumbleweed you are probably already benefiting from the result.

The second task was not exactly about fixing bugs present in the software, but about cleaning bugzilla from bug reports that were simply too old to be relevant any more. Like bugs affecting very old versions of openSUSE that cannot be reproduced in openSUSE 13.2 or Leap. We must be realistic about releases that are already out of support and the limited human resources we have. We closed around 80 of those ancient bugs.

no_country_for_old_bugs

So overall, we cleaned up around one hundred bugs from our queue. Still a long way to have a bug-free YaST, but undoubtedly a step in the right direction.

More coming

We are already working in the next sprint that will hopefully bring several new improvements for CASP, quite relevant progress in the storage reimplementation, several usability improvements, some bugfixes and even some news about this blog.

To make sure you don’t get bored while waiting we are also planning to finally publish the report about our visit to Euruko 2016.

Stay tuned!

Proprietary AMD/ATI Catalyst fglrx 15.12 rpms released for LEAP 42.2

October 15th, 2016 by

Warnings

There’s no warranties the drivers will work, for you!

If you are satisfied with the open-source radeon drivers, don’t risk to break your computer !

Still there will NEVER be a fglrx driver for recent kernel and xorg. So if one of those component change in Leap fglrx will be broken.

Actual situation

Since last december, AMD doesn’t published any update about fglrx so the version is still the 15.12.302 published. A few days ago our beloved Leap release manager Ludwig ask me by email, if there will be an available drivers for Leap 42.2.

Today, after hacking a bit the last Sebastian Siebert’s script I’ve been able to build the drivers for Leap 42.2 RC1, and the driver install fine, and xorg start on my HD5750 (but that’s all what I can tell).

I will rebuild the driver once Leap 42.2 will hit its final stage.

Repository

zypper ar -cfg -n FGLRX http://geeko.ioda.net/mirror/amd-fglrx/openSUSE_Leap_42.2/ FLGRX

zypper -v refresh -f FGLRX

zypper -v install fglrx64_amdcccle_SUSE422 fglrx64_core_SUSE422 fglrx64_graphics_SUSE422  fglrx64_opencl_SUSE422 fglrx64_xpic_SUSE422

Future

AMD has stopped any development for FGLRX, so it is already considered obsolete. But on the other side they make a lot of effort to bring radeon and amdgpu (the free and open source driver) to a decent performance level.

I don’t have that much usage anymore of my AMD gpu powered computer, and my HD5750 is now 8 years old already, so I can’t promise to be able to follow up with changes.

Cleanup

I removed all the obsoletes packages letting only the last one for each openSUSE version still available. Also the server has no more copy of openSUSE github artwork. If this missing to someone, don’t hesitate to ask.

Have fun

Highlights of YaST development sprint 25

September 28th, 2016 by

Another development sprint is over. Time flies! In our previous post we already reported about the branching of Tumbleweed and the upcoming releases and about the expected consequences: the landing of some cool features in a less conservative Tumbleweed.

We are still dedicating quite some effort to polish the upcoming stable releases (SLE12-SP2 and Leap 42.2), but in this sprint we finally found some time to play. Which is great because blogging about new features is more fun than doing it about bug fixes. 🙂

Importing Authorized Keys with AutoYaST

When logging in via SSH, public key authentication should be preferred over password authentication. Until now, the best way of setting up the required authorized_keys files in AutoYaST was using the files section.

However, that approach is tedious and error prone, as you need to make sure you set the correct owner, permissions, etc. Moreover you need to keep in sync the user definition (username and home directory) with the file definition.

AutoYaST now supports the specification of a set of public keys for each user with a pretty straightforward syntax:

<user>
  <username>suse<username>
  <authorized_keys config:type="list">
    <listentry>ssh-rsa your-public-key-1</listentry>
    <listentry>ssh-rsa your-public-key-2</listentry>
  <authorized_keys>
<user>

AutoYaST takes care of writing the files and setting the ownership and the proper permissions.

While documenting this new feature we realized the AutoYaST documentation about users management could be more detailed, which leads us to…

Improving the documentation

Usually developers love to create programs loaded with cool features but hate to write documentation. Fortunately there are people out there who enjoy writing documentation and bringing all those features to light. We have already mentioned in previous reports how grateful we are for having the SUSE documentation team polishing and publishing our documentation drafts and how open and straightforward the process is.

We updated the YaST documentation to include information about the installer self-update feature, which will debut in SUSE Linux Enterprise 12 SP2 and openSUSE Leap 42.2. As part of the same pull request and in the AutoYaST side, some additional improvements were made, including cleaning-up some duplicated information about SUSE registration.

On the other hand and as a consequence of the above mentioned new feature, the AutoYaST documentation regarding users management has been rewritten adding missing information like groups, user defaults and login settings.

All our pull requests are already merged in the doc-sle repository. At a later point in time, the SUSE documentation team will review and polish all the new content (including ours) and will publish an up-to-date version of the online documentation. If you don’t want to wait, you can easily generate an HTML or PDF version of the documentation including all the non-reviewed contributions just following the very simple instructions in the README file of the doc-sle repository.

Did we already mention we love the open source, programmer-friendly processes of the documentation team? 😉

Storage reimplementation: something you can touch

We promised news about the storage reimplementation and here they are. Our customized Tumbleweed image (labeled as NewStorage) in the storage-ng OBS repository can now perform some simple actions during installation and display the result to the user.

First of all, when proposing the timezone settings it will, as usual, check for MS Windows installations in the disk to guess if the hardware clock should be set to UTC. The news is that check is performed using the new storage stack, that offers more functionality in every sprint.

More important is that the installer will show the partitioning proposal calculated also using the new stack. As you can see in the screenshot below, the screen is way more simpler than the one you would find in a regular Tumbleweed. There are no buttons to change the settings or to run the expert partitioner yet. That doesn’t mean the functionality is not there, it’s simply that we prefer to focus first on modifying all the installer steps to use the new stack (what will enable us to use openQA) before refining every screen to add all options there.

The new partitioning proposal

Right now the system works only in disks containing a MS-DOS style partition table and will always propose a partition-based (no LVM) setup. That’s because we prefer to solve the hardest scenarios first. Using LVM and/or GPT partition tables is less challenging than the already supported scenario.

Reduce global warming by saving OBS build power

As you may know, we use the awesome Open Build Service (OBS) to generate both the YaST rpm packages and the openSUSE/SLE ISO images. Every time the source code of any component changes, OBS rebuilds that component and all the packages that depend on it.

Our beloved openSUSE and SLE release managers told us that there were several YaST packages that often triggered rebuild of other YaST packages, that triggered yet another rebuild, that triggered… you got the idea. 😉

The mentioned problem slows down the creation of new ISO images, interferes with the continuous integration process (specially visible in Tumbleweed) and wastes valuable OBS resources.

During this sprint we reduced the rebuild time of YaST by 30%. That’s for sure interesting, but knowing the details about how we did it could be even more interesting for many readers. We feared the explanation could be too complex and technical to fit into this report… which gives us just another opportunity for blogging. So expect an upcoming post including interesting technical stuff and crazy graphs like this one.

YaST dependencies graph

Some adjustment for the installer in the LiveCDs

One of the good things about working in open source is that sometimes the evolution of the projects you have created can surprise you. Quite some time ago, the YaST team dropped support for the live installer. It was simply too demanding to keep it alive while still doing our regular work (bug fixes and new features for YaST and the regular installer).

Recently the live installer was removed from Tumbleweed, the only system in which it was still available (after having been dropped in the past from stable openSUSE releases). One could have expected that somebody would decide then to step up and take the maintainership of the live installer.

But what actually happened was that Fabian Vogt decided to try a different approach we haven’t considered – adding the standard network installer to the LiveCDs images of Tumbleweed. He managed to make it work well enough and asked us for help to debug some problems. We fixed the initial problems by disabling the self-update functionality in the LiveCDs (it’s simply not designed to work on that scenario).

There are still quite some problems to be resolved to make everything work flawlessly, but if Fabian and others don’t give up, we will keep assisting them in order to bring the installation back to the LiveCDs… even in unexpected ways.

UI Designer

The YaST user interface is quite difficult to design and code. The main problem is that there is was no interactive UI designer where you could build a dialog or modify an existing one. Instead, you had to write new code or modify the existing code which creates and opens the dialog. Then, to see your changes you had to start the YaST module, go to the respective dialog and check its content. If it didn’t look like you intended, you had to close it, modify the code and start it again. And again… and again. Very annoying especially if the dialog is hidden deep in the module and you need to take several steps to get there.

During Hack Week 14, a project to improve the situation a bit was started. We already had a dialog spy which can be opened by Ctrl+Shift+Alt+Y keyboard shortcut, but that was read-only. You could only inspect the dialog tree and see the details of the selected widget but you could not change anything.

During that Hack Week project it was improved so it could edit the properties of the existing widgets, remove them or even add some new widgets. However the code was more or less a proof of concept than ready to be merged into the YaST UI and released to public. So we decided to finish it in this sprint.

As usual, it was harder than expected… but we made it and here is a short demo showing how it works and what you can do there:

The new UI designer in action

The new tool is still far from being perfect. The most obvious missing feature is that the dialog is changed in place and you cannot save or export you changes. After closing the dialog everything is lost. But it can still help to try things in the UI or make a quick prototype, specially when discussing solutions with interface designers. Hopefully we find some more time in the future to make it even better.

Storage reimplementation: encryption support

Although the partitioning proposal still does not support encryption or LVM, we implemented full LUKS (encryption) support in the underlying library (libstorage-ng). Together with the LVM support implemented in the previous sprint, that makes the new library already a valid replacement for the old libstorage in many situations and scenarios. Now it’s mainly a matter of switching from one version to another in every single YaST component, starting with the expert partitioner that we plan to start redesigning in the next sprint.

As usual, new features in the library are hard to illustrate, unless you accept the action diagrams as screenshots. In that case, here you can see the sequence of actions performed by the library when creating an encrypted home volume.

Creation of an encrypted home with libstorage-ng

Syncing keyboard layouts and console fonts in Leap and Tumbleweed

In parallel to our Scrum sprints, we have been for some time steady working, together with the openSUSE maintainers of X.Org and systemd, in redesigning how keyboard maps and console fonts are managed by YaST. Some changes were introduced in Tumbleweed some time ago but never made it to SLE (or Leap) because they needed more testing.

Recently Ludwig Nussel, the Leap’s release manager, decided that he wanted to sync 42.2 with Tumbleweed in that regard, using the new approach instead of the more conservative SLE one. Thus, we also invested quite some time coordinating again with Stefan Dirsch (X.Org) and Franck Bui (systemd) to push the changes just in time for the beta2 version of Leap 42.2… just in time to introduce bug#1000565, that was honored with its inclusion in the list of most annoying bugs in 42.2 Beta2.

The bright side is that a fix for the bug has already been provided (see bug report) and you can now finally test the new fonts and keyboard maps. Please, do so and provide feedback in order to get a properly localized Leap 42.2 release.

See you soon

As usual, this post was just a quick overview of the most interesting part of the sprint, because most people (including ourselves) don’t want to read about the boring part of the work, which is mainly fixing bugs.

The good news is that this time you will not have to wait another three weeks to read interesting stuff about YaST. As mentioned, we plan to publish a blog post about the reduction of the build time of YaST. And we will probably also publish a post about the visit of a YaST geecko to Euruko 2016.

So this time more than ever… stay tuned!

AMD Catalyst 15.12 for openSUSE – new makerpm-amd-script is available

March 17th, 2016 by

AMD has released the new AMD Catalyst 15.12 (Radeon Crimson Edition). My script replaces the existing packaging script with an updated packaging script. It supports up to Kernel 4.5. (Official support up to Kernel 3.19)

Important note: The driver does not work on openSUSE Tumbleweed. Unfortunately, the version of X-server is too new for the driver.

SHA1 is obsolete by now. The script used SHA256 for integrity of the downloaded files.

New Feature from packaging script:

  • systemd support

Resolved Issues:

  • [SWDEV-82980] Ubuntu 15.10 fails when building the .deb packages

Link: AMD Catalyst 15.12 Release Notes

Downloads:

Installation guide (English):
http://en.opensuse.org/SDB:AMD_fglrx#Building_the_rpm_yourself

Bruno Friedmann will build the new RPM packages in the fglrx repository. Stay tune!

If you find any issue with the driver. Don’t hesitate to contact me. I am in contact with AMD and can forward your issue to the right place. Feedback are welcome.

A report of your system is very helpful beside your feedback. You can generate it with the script:
su -c 'sh makerpm-amd-15.12.sh -ur'

Have a lot of fun!

Sebastian
openSUSE member / Official AMD Packaging Script Maintainer for openSUSE
German Blog: openSUSE – proprietären Grafik-Treiber AMD Catalyst 15.12 als RPM installieren

Announcing Li-f-e 42.1

December 21st, 2015 by

The best Linux distribution for education got a whole lot better, your Li-f-e(Linux for Education) takes a “Leap” to 42.1. openSUSE Education community is proud to present this latest edition based on openSUSE 42.1 with all the features, updates and bug fixes available on it till date. This effectively makes it the only enterprise grade long term supported(LTS) distribution for Education.

As with previous releases we have bundled a ton of softwares on this live DVD/USB specially packaged for education, along with the Plasma, GNOME and Mate Desktop Environments, full multimedia experience is also provided out of the box thanks to the Packman repositories. Only x86_64 architecture is supported, if you have a lot of machines that only support x86 then read on to find out how you can extend their Li-f-e.
(more…)

Proprietary AMD/ATI Catalyst fglrx rpms new release 15.11 (15.300.1025-1)

December 6th, 2015 by

Time to get an update for fglrx: the new release has been build for 13.1, 13.2, Leap 42.1, and tumbleweed 20151201

Tumbleweed beware : broken Xorg

It seems since the release, that a number of report of broken xorg with gdm, ssdm and so… segfaulting

You can still use the previous version hanging on the server. But I doubt it would work better

As soon, with Sebastian Siebert we have a patch for, I will republish a new build

Informations & bugreport Sebastian’s blog or lizards.o.o

The proposed drivers support kernel up to version 4.4

AMD release note available

Sebastien Siebert making script

Sebastian Siebert posts about fglrx

If you have any problems with the driver, don’t be afraid to report to Sebastian (German and English bugreports are gladly accepted).
he will try, as far as I am able to reproduce the bug. Together with the necessary system information, he will go directly to the right place at AMD to have the bug fixed in the next driver release.
Thank you very much, Sebastian.

See below what to do in case of troubles.

Or you can also ping him on irc (freespacer)

Debugging troubles

I recommend in case of trouble the use of his script which can collect the whole informations needed to help you. then you just have to issue a simple commande in console to collect all informations, you can review them, and finally transmit them.
Check the website to get the latest.

su -c 'sh makerpm-amd-15.11.sh -ur'
The system report 'amd-report.txt' was generated.                                                                                                          [ OK ]
Do you want to read the system report 'amd-report.txt' now? yes/no [y/n]: y
Are you sure to upload the above-named system report to sprunge.us? yes/no [y/n]: y

The report was uploaded to sprunge.us.
   The link is:  http://sprunge.us/eMEB

Copy paste the link in the comment zone of Sebastian post

All proudly distributed by openSUSE powered server and sponsored by Ioda-Net Sàrl