Tag Archives: linux

Some benchmarks for building an Eclipse project: Linux, Windows, macOS (M1)

Leave a reply

I want to share some benchmarks for building an Eclipse project with Maven/Tycho. I built the project on Linux, Windows, and macOS (in particular, on my Mac Air M1 2020).

I stressed that the macOS environment is an M1, the new Arm-based chip from Apple, which is said to be quite efficient and performant. However, I also used an older Mac Air 2016, so no M1 here.

Concerning Linux and Windows, I used an LG GRAM 16. Both are good machines with powerful processors, 16Gig RAM, and NVMe disk. I ran the same build on a few Linux distributions (Arch, Ubuntu, Fedora) and the results concerning Linux are basically the same.

The Eclipse project is Edelta, https://github.com/LorenzoBettini/edelta, “An Xtext DSL for EMF metamodels refactoring and migration.” It’s quite a complex project with an extensive test suite: unit tests, integration tests (JUnit Plug-in tests), and SWTBot tests. The project also includes a custom Maven plugin, which is also tested. Yes, I’m test-addicted 😉

To avoid the Internet connection influencing the benchmarks, I first run the Maven build once so that Maven downloads and caches all the dependencies. Moreover, I first mirrored the p2 Eclipse artifacts so that the target platform definition is resolved against the local mirror.

LG GRAM 16 (LINUX) Total time: 03:25 min

LG GRAM 16 (WINDOWS) Total time: 04:20 min

MACBOOK AIR M1 Total time: 02:49 min

MACBOOK AIR 2016 Total time: 08:17 min

I must say I was impressed by the best performance shown by the M1 processor!

As expected, on the same machine, Windows performs worse than Linux 🙂

The performance of the old Mac Air is not impressive, but that’s an old machine.

Just for fun, I ran the same build on the PineBook Pro (which runs Manjaro KDE): Total time: 21:27 min. Of course, that’s not usable for intensive development 😉

My notes on the Grub incident in Linux Arch

Leave a reply

In this post, I’d like to share my thoughts and the solution I adopted to circumvent the terrible incident that happened a few days ago with grub in Linux Arch. The bug affected all Arch-based distributions (not Manjaro, which is not really Arch-based, but for sure EndeavourOS and Arch itself).

You can find the details of the problem online:

If you’re reading this, you’re already aware of the bug, and maybe you’ve been hit by it. Thus, I’m not going into the details of the issue itself.

First of all, although I’ve been enjoying Arch for some time, I must say I was not pleased with the way the Arch team handled the problem (it’s really a severe issue, and having bootable snapshots does not save you from that if you don’t take countermeasures to avoid it or to fix the problem, once you’ve already been hit). After a few days, the Arch team closed the bug, blaming other Arch-based distros because they have hooks that update the grub configuration during package upgrades. The Arch team says that if you use grub the “Arch way,” you have no problem. However, if I understand the details of the issue correctly, as soon as you have a software update that installs a new kernel version, it will also update the grub configuration, and the bug will hit you. (update: no, that’s not the case in Arch, since the kernel image file always has the same name and does not have the version in the name).

Moreover, I have hooks to take Timeshift snapshots before a system upgrade and to generate grub entries to boot the snapshots. Isn’t this the “Arch way”? For sure, it’s a safety mechanism widely adopted. After all, Linux is about choice! 🙂 (update: since, as said above, the kernel image file name is always the same, even, for example, if you update from version 5.19.4 to 5.19.5, having bootable snapshots is even more crucial in Arch).

The issue is said to be fixed by showing a warning during the upgrade of the grub package saying:

Seriously?! That’s just a warning, which is likely to go unnoticed. That’s not a fix; it’s a patch (In the true sense of the word).

However, let’s say that you note that warning and do what it says.

You will have no problem on the next boot, but I found these huge drawbacks (at least for me):

  • The grub installation of the current Arch distribution will replace the default UEFI boot configuration. Of course, you can reorder the boot configurations from the UEFI setup, but that’s an unwanted side-effect if you had already configured things the way you wanted;
  • Related to the previous point, if you have another grub configuration (e.g., of another distro) to boot other distros on the same machine (see, e.g., my blog post on this mechanism), including this Arch installation, it will not work anymore. That’s expected because the other grub does not know anything about the new grub configuration anyway.

Moreover, in my humble opinion, what the grub developers have done in this new version does not make much sense: why do you want to execute a command to generate a boot entry dynamically?! Perform that choice when generating the grub configuration once and for all!

If you’re curious, that’s the difference after the grub update in the file used to generate the culprit entry in the grub.cfg (old file on the left, new file on the right):

So, after restoring my computer with arch-chroot (because I was hit by the bug), instead of their proposed “solution,” I ignored the package update by adding this line to /etc/pacman.conf:

This way, I won’t be hit by the bug again (until, hopefully, grub developers will revert that implementation decision, and Arch developers will provide a new grub package)

Then, I realized that, in the end, I don’t need a grub entry to enter UEFI! I can enter UEFI when starting the computer with a key combination (not to mention that, as I said above, my main grub is not the Arch one, but the Fedora one, as shown in my blog post)! Thus, I updated the grub package (of course, after removing the IgnorePkg entry shown above). Before rebooting (that’s crucial), instead of following the suggestion shown during the upgrade, I simply removed the grub file to generate the culprit configuration for UEFI in the grub.cfg (and, of course, I regenerated the grub configuration afterward):

That’s all! Your grub.cfg will not have an entry for entering UEFI anymore. I don’t need that.

Of course, you have to repeat these operations if Arch releases an update to the grub package without fixing the problem. However, I prefer this solution, which does not have the abovementioned drawbacks.

A first look at Crystal Linux

Leave a reply

Crystal Linux is a brand new Linux distro based on Arch. Although it has not been officially released, you can already download an ISO and try that in its current development state.

Quoting from its home page:

Crystal Linux is a brand new Arch Linux based distribution.
Powerful and easy to use.

And, in particular

Why Crystal?
What’s so different about Crystal compared to other distributions?

  • Easy to use package manager
  • Beginner Friendly
  • Easy Btrfs snapshots
  • Easy to install

Since I’ve been using Arch and Arch-based (mainly EndeavourOS) distros for a while, I decided to try it, even if it’s not officially released, to briefly evaluate its current development status.

I downloaded the ISO crystal-live-08-20-22-07-55-x86_64.iso and tested it on a VirtualBox VM.

After the BOOT, you’re greeted by the GNOME desktop. I searched for the installer. I thought I’d find it in the dock, but that was not the case. I searched for an “install” string in the GNOME Activities. Again I got no results. Then, I recalled that Crystal names its installer Jade, and that’s how I found it.

The installer provides several screens to specify the typical information of a Linux installation. I would say that it does that in a very clear way. Here are a few screenshots of the pages of the installer:

Since it’s based on Arch, Crystal can provide a good selection of desktop environments. However, you can select only one:

Another interesting section is the next one, where you can enable Timeshift (recall from the introduction that “Easy Btrfs snapshots” is one of the advertised features):

Regarding partitioning, the Crystal Linux installer shows its “work-in-progress” state: you can only select the disk for the installation and nothing else. As noted in the dialog, it will wipe the whole disk and automatically partition it. Nothing is said about the file system, but I guess it will be a Btrfs filesystem (Spoiler: it is):

However, they are still working on the partitioning program:

Then, you get the summary for a final review. When you’re ready, start the installation:

The installer shows the installation log, which is excellent and helpful. Unfortunately, when the installation finishes, the log is not available anymore. It would be good to have access to that window still, to copy the log somewhere for a later examination.

OK, let’s reboot and see the distribution in action.

Crystal Linux should come with a customized GNOME desktop, Onyx, but that’s not installed by default. So you get a vanilla GNOME session (just like in Arch) with only a few programs installed (so, in this respect, it’s not bloated):

You don’t even have a Text Editor with a GUI (but at least you have Gnome Tweaks):

Of course, you can install whatever you want with pacman or with the Crystal custom AUR helper, amethyst, but beware: the command to run it’s ame; it took me some time to discover that 🙂 The AUR helper amethyst looks nice. Just like “yay”, you run it, and it also executes an update (“-Syu”). Besides that, the command line arguments are the same as pacman and yay, at least, that’s my first impression. Just like yay, amethyst also updates the AUR packages. Moreover, it is also configured to execute “pacdiff” at the end:

Unfortunately, the keyboard layout and the time zone I specified during the installation was not considered. I had to set them again in the GNOME settings.

The default shell is ZSH, which I like. However, it’s not configured at all, with any default. Thus, the first time you open a terminal, you have to do that yourself:

Let’s see how the installer created the partitions:

So we have BTRFS, with the two standard volumes @ and @home, which are suitable for Timeshift. No SWAP partition (swap is not configured at all). Moreover, there’s no compression on the BTRFS subvolumes, while I usually prefer to have compression.

There are also other subvolumes (I guess to exclude from snapshots the virtual machines you might create with Kvm):

Concerning repositories, Crystal is similar to EndeavourOS: it relies on Arch repositories and adds just a few custom repositories that you can also browse at https://repo.getcryst.al/. You can see that there are not so many packages there. Besides the Crystal programs, you also find a few packages you typically install from AUR, like timeshift and timeshift-autosnap.

This allows me to talk about the “Easy Btrfs snapshots.” The Crystal installer has already set up timeshift, timeshift-autosnap (which creates snapshots before package upgrade), and grub-btrfs (which adds “btrfs snapshots” to the grub menu to boot your system on a “snapshot” from the grub menu). (I wrote about these 3 tools previously).

For example, that’s what you see when you perform an upgrade (note the automatic creation of the snapshot and the creation of the grub menus for the snapshot):

In fact, the next time you boot, you also get the grub entry for the snapshots:

Summary

I liked a few things and a few things I did not like, but I enjoyed the installation, and, in general, I have an excellent impression of this new distribution. It’s not for beginners, but it’s easy for users who have already tried a few Linux distributions.

In any case, I found Crystal promising, and I really look forward to the official ISO release!

Ubuntu Jammy and Nvidia

Leave a reply

I installed Ubuntu Jammy a few months ago, when it was released, on a desktop computer. It was a terrible experience: crashes now and then, and the Wayland session was not working well on some applications. After a week, I uninstalled it immediately and switched to Fedora on that computer.

Then, a new point release came out, 22.04.1. I heard that, in the meantime, many bugs were fixed. I decided to give it another chance. In particular, I decided to install that on an older computer, a Dell m3800, because it comes with an integrated video card and a “terrible” Nvidia card, which is well-known to give headaches to Linux users 😉 In the past, I used Bumblebee on such a computer to keep the power consumption low. However, Bumblebee gave me problems in the past, and I seem to understand that it is not actively maintained anymore. So, let’s try a fresh installation of Ubuntu Jammy and see how it deals with the Nvidia card.

The installation went fine and smoothly. In the end, when rebooting into the new installation, I could verify that the Nvidia drivers were installed automatically and working correctly (meaning no crashes or other issues)!

Since it uses Nvidia, Ubuntu does not show any gear icon on the login screen to switch the graphical session: it’s X11 automatically, and you cannot change it. I guess that’s done because Nvidia does not work correctly in a Wayland session.

Another nice thing is that it detected my huge screen resolution, 3200×1800, and automatically enabled 200% scaling. However, I seem to remember that this is a feature of Gnome itself.

Having said that, I found using the Nvidia drivers really a problem: the computer gets immediately hot, the fans get pretty noisy, and it looks like the battery does not entirely charge even when connected to the AC. It also looks like the battery goes down a bit, even when connected to the AC.

I opened the Nvidia settings. However, it looks like Hybrid mode will not work with this version of the driver.

Of course, when running on battery, the power consumption is too much, and the battery would drain in less than two hours, as reported by powertop. That’s the reported consumption (even when the computer is idle):

33.6 W of power consumption is too much.

Since I’m not using graphical intensive software, I tried to switch to the integrated Intel graphics card. You cannot select this choice in the Nvidia settings because it’s greyed out (you can find many posts on the Internet reporting this problem without a solution). However, we can use the command line:

Let’s reboot, and now the gear icon appears, and you can select the Wayland session since the Nvidia card is not used. By the way, Nvidia settings still have the greyed-out entry for the integrated card, but at least it recognized that’s the one selected:

The power consumption on battery is better (about 15 W, i.e., half), but I seem to understand that the Nvidia card itself it’s still using power since it’s not completely turned off:

At least the computer is not that hot anymore, and the battery seems to charge correctly.

Let’s try to use a tool to turn the Nvidia card off: EnvyControl.

Let’s install that:

As noted here, there are a few additional steps when installing EnvyControl in Ubuntu:

We could use the envycontrol command to set the card, but let’s now install a Gnome extension with a GUI for the switch: GPU profile selector.

Let’s also activate the extension.

Let’s reboot, and now, in the Gnome dropdown menu about the battery and power profile, we also have the menu entries to select the card:

Remember that we had to set the “on-demand” mode for installing EnvyControl. That’s why the “Hybrid” entry is selected.

Let’s select “Integrated” and reboot; now, the Nvidia card should be off.

The power consumption on battery is even better now (13.8 W):

If we switch to the Wayland session, it even improves a bit:

I find this installation usable, and I saw no issues for the moment.

That’s all for now!

Cropping images with KDE Gwenview

Leave a reply

I’ve always been using GIMP for cropping images (for articles, books, etc.), and it works pretty well, but I never thought that you could do the same much quicker with the default image viewer of KDE, Gwenview. In KDE Plasma, I’ve always used Gwenview as an image viewer, but it’s much more:

Gwenview is a fast and easy to use image viewer by KDE, ideal for browsing and displaying a collection of images.

Features:

  • Supports simple image manipulations: rotate, mirror, flip, and resize
  • Supports basic file management actions such as copy, move, delete, and others
  • …

In fact, you just open an image with Gwenview:

Press “Show Editing Tools” and then select “Crop” from the palette:

The box for cropping is ready to be resized:

When you’re done, just press ENTER:

and the image is ready to be saved!

With GIMP, the number of steps is much more significant.

Moreover, with Gwenview, once you cropped and saved the current image, you just press the right arrow to go to the next one. Much faster!

Installing Arch Linux with the archinstall script

Leave a reply

After using EndeavourOS for a while, I also decided to try Arch itself!

I did not feel ready to go through the manual procedure (but I’ll do that someday), and I heard about the archinstall script, which comes in the Arch official ISO. After watching a few videos and reading a few blog posts, I tried that myself, first on a VM and then on a real machine.

I can anticipate that I liked this installation procedure. Still, it is not perfectly usable in a multi-boot environment, as I’ll say near the end of this post, where I summarize my experience with the installation.

In this post, I’ll first show an installation on a VM (and I suggest you try that one), then briefly describe the installation on a real machine.

If you want to try that, you have to download an official Arch ISO. Once you booted into the live ISO, you must connect to the Internet as described in the official documentation.

Now, it’s time to run the installer:

And you have access to the main installation menu:

You see, it’s easy to use, especially if you’re familiar with other Linux installation programs.

The first three entries are easy to deal with.

The fourth, “Select harddrives,” requires some care because it’s where you deal with your disk! You have to select the correct drive. In my case:

Since I’m on a VM, I’ll simply choose to wipe everything on that drive and let the installer handle the partitioning automatically. That’s easy in a VM, and that’s what I’ve seen in all demos on the web (but, as I’ll show later, things are more complicated in a multi-boot environment):

Then, you choose the filesystem EXT4, BTRFS, etc.

For the bootloader, I chose GRUB, which is the one I’ve always used.

Concerning the user accounts, it’s best NOT to set a root password: it’s better to create a user with administrator rights (so that you later rely on the good ol’ “sudo”). An interesting feature of this installation procedure is that it lets you create as many users as possible. On the contrary, typically, other Linux installations only allow you to create a single user.

The other interesting menu entry is the one to choose the profile. I’m choosing “desktop”:

And in particular, I’m choosing GNOME (you see that you have plenty of choices):

Moreover, you can select the graphic drivers:

And the kernels:

Since, for the moment, you could just select from a predefined set of choices, you are given a chance to manually specify additional packages to install (but you have to know them by their name). In this example, I’m installing “firefox”:

It’s also crucial to configure the network for the installed system. If you use GNOME or KDE, I’d say that it’s best to choose “NetworkManager”:

Once you’ve done with all the menu entries, before starting the actual installation, you’re given a chance to save these configurations, which is helpful if you want to use the same configurations on other machines or to do some further customizations:

Now, it’s time to choose “Install”; a countdown starts to abort the installation in case you just remembered you’ve done something wrong:

The installation starts and in a few minutes (where a few packages will be downloaded)…

…you should get to the end of the installation, where, if you want, you can also tweak the installed system before rebooting:

If you choose “no”, you’re back to the live environment:

And you can now reboot to (hopefully) enjoy your installation:

Having installed GNOME, I’m presented with a few options, and I choose the first one, that is, GNOME on Wayland:

You see that the GNOME installation is a vanilla one.

OK, that was a VM, and it was straightforward to install Arch with the installation script. It’s also easy if you plan to install ONLY Arch on your computer (by wiping all the rest).

Things are not working completely fine if you want to install it on a computer with other Linux installations, which you want to keep and be able to boot into. In this case, of course, you cannot wipe the selected hard drive and must do manual partitioning. The installation script is still helpful in that respect (I’m not showing anything in this blog post), but you must be aware of a small problem.

In fact, due to an issue, which, at the time of writing, is still open, specifying to mount an existing EFI partition into “/boot/efi“, which, as far as I know, it’s the standard mount point in most distributions (every distribution I know at least). You must specify a mount point “/boot” and that must be a boot partition (with the boot flag). Of course, that’s what I’ve done myself; I specified the exiting EFI partition (the first partition of the installation drive in my computer) for the mount point “/boot”. However, the installer will treat that directory as if it was /boot/efi in other installations. Thus, it will copy the booting files directly there. As a sad result, the Arch installation will not be detected when rebooted. You will only see an existing installation’s GRUB menu (and running os-prober from an existing installation does not seem to help). Thus, you end up with the Arch installation that you cannot boot.

The only solution I found to boot into the Arch system was to apply the mechanisms shown in my other post and configure the Fedora GRUB with an entry pointing directly to the EFI partition, i.e., according to the post mentioned above, something like the following (remember, on my computer the EFI partition is the first one, and I installed Arch on the partition 13):

Or even like that (i.e., without specifying the root partition of the Arch installation at all and just relying on the grub that the installer created directly on the EFI partition):

Besides this problem, the installation script archinstall is really interesting and still under development.

Locate and BTRFS

1 Reply

I’ve always been using the locate command (provided by the mlocate package or by the new plocate package), which quickly searches for files and directories by their names. The command relies on the database built by the command updatedb (which should be run periodically, e.g., by enabling the plocate-updatedb.timer service for plocate).

Unfortunately, by default, it does go well with the BTRFS filesystem and its subvolumes (see, e.g., this bug), resulting in empty results for all searches basically.

Fortunately, the solution is quite simple:

  1. edit the file /etc/updatedb.conf
  2. replace PRUNE_BIND_MOUNTS = “yes” with PRUNE_BIND_MOUNTS = “no”
  3. save, exit and re-run updatedb

Then, you can enjoy locate’s search results 🙂

Linux EndeavourOS Artemis Review

Leave a reply

I have already blogged about EndeavourOS, which I use most of the time on a few laptops. Since EndeavourOS, based on Arch, is a rolling release, I update it almost daily and don’t need to install it from scratch when a new release comes out, like Artemis, which was released a few days ago. However, since I wanted to switch from the EXT4 file system to BTRFS (since I started to experiment with this file system and its snapshot capabilities), I took the chance to try this new release by installing it from scratch (of course, using BTRFS this time).

I’ll first go through the installation, but I can anticipate that, once again, I’m impressed by EndeavourOS. This installation feels really fast, maybe due to BTRFS or the new kernel (instead of the LTS kernel, I now use the latest one provided by the distribution) or both. Most of all, EndeavourOS is pure Arch but with outstanding defaults. Indeed, the KDE and GNOME environments are vanilla ones.

Installation

As usual, the first thing to do, once booted in the live environment, which in this case is XFCE, is set up the network connection. You might also want to change the keyboard layout (Disable system defaults and install your layout, in my case, it’s the Italian layout):

Then, let’s update the mirrors (typically by selecting your state) and start the installer.

I choose the “Online” method because I want to install KDE Plasma instead of Xfce.

You have to wait a few seconds (or about a minute) for the installer to download the modules (I always prefer to install any operating systems in English):

Maybe, due to a bug, the location has been found successfully, but the English version proposed is not the right one, so I have to change it to Americ English again:

After setting the keyboard layout (this time for the installed system), it’s time for partitioning.

Since on this computer I have a few Linux installations, including the old version of EndeavourOS I’m going to replace, I choose Manual partitioning (and not “Replace a partition” because that would keep the same file system type, EXT4, while I want to switch to BTRFS).

I Edit the partition, select “Format” (otherwise, I cannot change the File system type), modify the File system (BTRFS), and specify the mount point.

Before going on, we must specify to mount the EFI partition (into /boot/efi) without formatting it and ensure the “boot” flag is selected. This way, the installer can properly install GRUB.

WARNING: on another computer, the installer complained that I did not select a boot partition with at least 300Mb (mine was just 200Mb). Since I knew there was enough space in that partition, I ignored the warning, and the installation went fine.

As for the desktop, I select Plasma.

And then, we can select the single packages. Note that, different from my previous review, you choose the packages after selecting the desktop, so that a few packages, in particular, the ones of the chosen desktop, have already been selected:

In this blog post, I’m not going to install GNOME besides KDE, but I also select the “Printing-Support” and the “Support for HP Printer/Scanner” checkboxes.

As usual, then you have to specify your user’s details, and then it’s time to take a look at the summary:

Let’s start the installation. It might take a few minutes because it’s an “Online” installation, so it has to download several packages.

You can press “Toggle log” during the installation to see the installer’s log.

When it’s done, it’s time to restart the computer.

First impressions

As in the previous blog post, I must note that the Discover icon is still in the taskbar, though the software manager Discover is not installed at all.

Maybe because I selected KDE Plasma only, the Wayland session has not been installed, while in my previous post, I installed both GNOME and KDE. However, I just had to run this command:

And then I could enjoy Plasma also in Wayland, which seems to be pretty usable.

A nice addition is a firewall applet in the taskbar

Here’s neofetch:

Printing support is excellent! I connected an old USB HP Deskjet printer, and I got these notifications from Plasma:

The printer has been automatically installed correctly (I only had to configure a few things like paper size and color mode).

Setting up my Google accounts (drive access and calendar) is as cumbersome as always, but I did not experience any problems once I finished.

Power consumption on the battery is also excellent, but that was true in the past, so nothing changed in that respect.

All in all, this distribution keeps on being awesome.

As I initially anticipated, you have Arch and its vanilla desktop environments, but with useful and reasonable defaults. Moreover, the installation is effortless! 🙂

Enabling Hibernation in Arch-based distros

Leave a reply

I had already posted about enabling hibernation in Linux, particularly in Ubuntu.

Thanks to the script hibernator, the procedure is much more straightforward in Arch-based distros (including Manjaro and EndeavourOS).

First of all, make sure you first install the package update-grub.

The package hibernator is available from AUR, but currently, there’s a problem with the build instructions. Thus, we must install it manually. There’s no need to install the script anywhere: it’s just a matter of cloning the script from GitHub and running it once:

Then, before running the script (as a superuser), we must decide where we want the suspend-to-disk to take place: either in a swap partition or in a swap file (please keep in mind that currently, the script cannot handle hibernation into a swap file of a BTRFS partition).

If we want a swap file, the script can create that for us, and we can also specify the size of the swap file. Here’s the quote from the home page of the project

Hibernator accepts desired size of swapfile as arguments. Running hibernator 2G creates 2 Gb swapfil, hibernator 1000M creates 1000 Mb swapfile. The script defaults to 4G if no arguments are given.

If we want a swap partition, we must ensure the partition is already in place and that our /etc/fstab already refers to that (i.e., it mounts it appropriately).

When we’re ready, we just run the script with sudo. The script will update the GRUB command line with the resume option and a resume hook to /etc/mkinitcpio.conf. Finally, it will update the GRUB configuration (with update-grub, that is why you need to install this package beforehand).

Here’s an example of the output:

And that’s all: we just reboot, and hibernation is ready to be used!

Before rebooting, you might want to check that the GRUB_CMDLINE_LINUX_DEFAULT variable in /etc/default/grub has been set with a valid resume entry (if you rely on a swap partition, which has been properly specified in /etc/fstab, it should contain a reference to the UUID of the swap partition):

You can test that, as usual, with

You may also want to refer to the older post enabling hibernation in Linux, particularly in Ubuntu, for other mechanisms related to hibernation, like suspend and then hibernate.

Timeshift and grub-btrfs in Linux Arch

Leave a reply

After looking at the very nice videos of Stephen’s Tech Talks, in particular, this one https://www.youtube.com/watch?v=6wUtRkEWBwE, I decided to try to set up Timeshift, Timeshift-autosnap, and grub-btrfs in my Linux Arch installation, where I’m using BTRFS as the filesystem. These three packages allow you to have a timeshift snapshot automatically created each time you update your system; moreover, a new grub entry is automatically generated to boot into a specific snapshot.

The video mentioned above is handy, but unfortunately, some recent changes in Timeshift itself broke the behavior of the two other packages. In this post, I’ll try to show how to fix the problem and go back to a working behavior. I’ll also show an experiment using the snapshots so that, hopefully, it’s clear what’s going on in the presence of such snapshots and how to use them in case you want to revert your system.

First of all, let’s install timeshift and timeshift-autosnap (the latter depends on the former, and they are both available from AUR; I’m using the AUR helper yay here):

The first problem is that timeshift has recently changed the strategy for creating snapshots. Instead of creating them in /run/timeshift/backup/timeshift-btrfs/snapshots, it now creates them in /run/timeshift/<PID>/backup/timeshift-btrfs/snapshots, where <PID> is the PID of the Timeshift process. Each time you run Timeshift, the directory will be different, breaking grub-btrfs (which expects to find the snapshots always in the same directory).

Fortunately, there’s a workaround: we add an entry to /etc/fstab in order to mount explicitly the path /run/timeshift/backup/timeshift-btrfs/snapshots:

where, of course, <UUID> has to be replaced with the same UUID of the physical disk partition.

Reboot, and then Timeshift will also put the snapshot in that directory (besides the one with the PID, as mentioned above). You can try to create a snapshot to verify that (this also allows us to use the Timeshift wizard so that we specify to create BTRFS snapshots).

Let’s make sure the mount point is active (and note the unit name)

Let’s now install grub-btrfs

We need to configure that to monitor the Timeshift snapshot directory instead of the default one (/.snapshots).

The file contents

should be replaced with

Let’s reload and re-enable the monitoring service:

If we have already created a few snapshots, we can run update-grub (or, if you have not installed the package update-grub, use the command “grub-mkconfig -o /boot/grub/grub.cfg”) and verify that new grub entries are created for the found snapshots:

We can also restart the system and prove that we can access the GRUB submenu with the generated entries for the snapshots.

IMPORTANT: If you have several Linux distributions on your computer and you use a multiboot system like the one I blogged about, and this distribution is not the main one, you will have to manually tweak the entry in your main distribution’s GRUB menu. See the linked blog post near the end.

Some experiments

Let’s do some experiments with this configuration.

Here’s the kernel I’m currently running:

I’m updating the system (I’m skipping some output below, and you can ignore the “stale mount” errors):

So it created a snapshot before updating the system (in particular, it installed a new kernel version). Let’s reboot and verify we are running the new kernel (5.18.8 instead of 5.18.7):

Let’s reboot and select from GRUB the latest snapshot (remember, the one before applying the upgrade), so timeshift-btrfs/snapshots/2022-07-02_15-35-53 (snapshots are presented in the grub submenu from the most recent to the oldest one). We do that by pretending that the update broke the system (it’s not the case), and we want to get back to a working system before the update we have just performed.

You see that the “Authentication Required” dialog greets us, and in the background, you can see the notification that we “booted into Timeshift Snapshot, please restore the snapshot”:

The password is required because it’s trying to run Timeshift:

In the screenshot, you can see that we are now using the older kernel since we booted in that snapshot where the update has not yet been performed. We have to restore the snapshot manually; otherwise, on the next boot, we’ll get back to the updated system version and not in the snapshot anymore.

So, let’s restore the snapshot:

You see, Timeshift has just created another snapshot ([LIVE]). We now reboot normally (that is, using the main grub entry, NOT the snapshot entries).

Once rebooted normally, we can verify again that we are running the old kernel:

Let’s have a look at Timeshift, and we can see the last snapshot is an effective one, not a LIVE one:

Yes, we are now in a system where the update above has never been applied.

Let’s try to rerun the update command (we don’t effectively execute the update, it’s just an experiment):

Why? Because the snapshot had been created automatically by timeshift-autosnap before applying the updates while the package manager was running, its lock is still there.

Let’s remove the lock and try to rerun the update:

The output is similar to the one shown above (unless there are even more new updates in the meantime, which might happen in a rolling release), but something is missing:

Why? Because the downloaded packages in the cache are NOT part of the saved snapshot, they are still present in the current system, even though we restored the snapshot. Why are the cached packages still there, but the lock has been restored with the snapshot? That’s due to the way subvolumes are specified in the /etc/fstab:

You see, the cache of downloaded packages and the logs are NOT part of the snapshots, while /var/lib (including the pacman lock) is part of the snapshots.

Let’s now revert the snapshot: we select the one with “Before restoring…”.

Again, we are now in a LIVE situation, and Timeshift tells us again to reboot to make it effective.

Let’s reboot (normally, by using the main grub entry).

We’re back to the updated system, and there’s nothing to update (again, unless new updates have been made available in the meantime):

If we’re happy with the updated system, we can also remove those two snapshots (remember that grub-btrfs monitors the snapshots so that it will update its grub submenu entries):

I hope you find this blog post helpful, and I hope it complements the wonderful video of Stephen’s Tech Talks mentioned above.