Buster to bullseye upgrade: Worth it or not?

[local10]

Hi,

Am considering upgrading from buster to bullseye and looking for some feedback from those who have already done so. Was it worth it, in your opinion? Anything you like in particular about bullseye? Any problems/issues or degradation of performance/functionality/features, especially related to the KDE and its apps?

I'm asking this because something upgrading just isn't worth it, it's better perhaps to wait for the next release, like for example when KDE 3.5 was replaced with the initial KDE 4.x, that upgrade created so many problems (for me) while adding very little in terms of benefits.

Thanks

[Greg Wooledge]

On Mon, Oct 11, 2021 at 02:59:35PM +0200, local10 wrote:
> Am considering upgrading from buster to bullseye and looking for some feedback from those who have already done so. Was it worth it, in your opinion? Anything you like in particular about bullseye? Any problems/issues or degradation of performance/functionality/features, especially related to the KDE and its apps?

You mention KDE. So, let me share this little story with you.

We (i.e. my employer) have some Debian workstations that are set up to
run a commercial data review/analysis software suite. The vendor of this
particular software suite started out on HP-UX, and later migrated their
support platform to Linux. They use a variant of CentOS, and they use KDE.

I set up some Debian 9 (the stable version at that time) workstations, and
managed to get them up and running the vendor's software suite. Since the
vendor used KDE on their supported workstations, I used KDE as well, in
order to minimize confusion for the users.

The workflow model was pretty straightforward: data were collected by
one data acquisition system (running the vendor's Linux variant), and
reviewed by trained users sitting at the data review workstations (running
Debian 9).

Then the pandemic happened.

Did I mention that this is *medical* data? It's medical data. The users
are physicians, or highly trained medical specialists.

So, in the early stages of the Covid-19 pandemic, access to the physical
workstations was severely limited. People were not allowed to come to
work in person except in special cases (e.g. the people who actually
guide the patient into the room, hook up the electrodes, etc.). All of
the people who were supposed to review the data were stuck working from
home.

Skipping over a *whole* lot of detail, what we ended up doing was setting
up a VNC server on each of the data review workstations. The users who
are reviewing the data from home connect to that, and run the vendor's
software suite inside a VNC session.

The problems did not start immediately.

Periodically, for reasons I could not initially pinpoint, one of the VNC
sessions would simply stop working. Stopping and restarting it did not
help. The only thing I could do was reboot the entire workstation.

After several months of this, running every command I could think of,
doing Google searches, etc., this is what I came up with:

It seems that when KDE's SDDM (display manager) locks the screen, and
a user unlocks the screen, there are two different things that can
happen. Either the screen unlocks normally, and the user resumes their
KDE session; or, SDDM launches a whole *new* KDE session, and now the
user is logged in twice.

So, as near as I can tell, every once in a while (increasingly often
as users were allowed back on site), someone would sit down at one of
the workstatiosn, try to unlock the screen, and would be thrown into
a whole new session. They would be logged in twice on the console, and
running two different X sessions, one on display :0, and the other on
display :1. Sometimes this would show up in "who" output. and sometimes
it wouldn't. Sometimes I'd see a /tmp/.X11-unix/X1 socket owned by
root instead of the VNC session's user. Sometimes I wouldn't.

Whenever this happened, the VNC session that was trying to use display :1
would fail. Only rebooting would fix it.

My Google searches told me that this was considered a "feature" in
SDDM, but was annoying for many people, because they kept tripping it
accidentally. For them, it wasn't a disaster -- they would be logged
in twice, and they'd struggle to figure out that they could get back to
their original session by hitting Ctrl-Alt-F7 or whatever.

For *us*, it meant someone was completely locked out, and I had to
intervene to fix it. This was more than simply annoying.

As it turns out, there's a configuration option to control this feature
in SDDM. But only in *new* versions of SDDM. It didn't exist in the
version shipped with Debian 9. And it didn't exist in the version
of SDDM shipped with Debian 10 either. Apparently it *does* exist in
the version shipped with Debian 11, but it's undocumented. It's also
supposedly enabled (fixed) by default.

So, the point of all this is that upgrading to Debian 11 helped us work
around a major design flaw in SDDM (KDE's display manager). It hasn't
caused me any problems yet (although having to rework NIS was a small
hassle).

As always, please read the release notes, and also check
<https://wiki.debian.org/NewInBullseye> for the latest set of known
issues reported by the user community.

[Andrei POPESCU]

On Lu, 11 oct 21, 14:59:35, local10 wrote:
> Hi,
>
> Am considering upgrading from buster to bullseye and looking for some
> feedback from those who have already done so. Was it worth it, in your
> opinion? Anything you like in particular about bullseye? Any
> problems/issues or degradation of performance/functionality/features,
> especially related to the KDE and its apps?

One machine here is running bullseye (fresh install), and so far it
seems to be fine, though I didn't have the chance to push it to its full
potential / intended use.

Since the hardware is quite new it might not even work very well with
buster (if at all).

The other systems under my care will be upgraded soonish, unless they
are completely de-commissioned / re-purposed (for unrelated reasons).

With the upgrade to bullseye I would get rid of a few backports, but
nothing essential.

This may or may not be relevant for your use, more details from your
side could result in more accurate assessments.

> I'm asking this because something upgrading just isn't worth it, it's
> better perhaps to wait for the next release, like for example when KDE
> 3.5 was replaced with the initial KDE 4.x, that upgrade created so
> many problems (for me) while adding very little in terms of benefits.

Apparently that particular (dist-)upgrade was indeed painful for many
KDE users[1]. The rest of the system was fine though, as far as I
recall.

On the other hand regular security support for buster will end about one
year before the release of bookworm. Are you fine with (only) LTS
support?

https://www.debian.org/lts/

[1] It might have contributed to my brother eventually giving up on
dual-booting Debian on his desktop.

Kind regards,
Andrei
--
http://wiki.debian.org/FAQsFromDebianUser

[Dan Ritter]

local10 wrote:
> Hi,
>
> Am considering upgrading from buster to bullseye and looking for some feedback from those who have already done so. Was it worth it, in your opinion? Anything you like in particular about bullseye? Any problems/issues or degradation of performance/functionality/features, especially related to the KDE and its apps?

I have upgraded 8 machines at home and had no negatives at all.

Several of those are desktops, but none run KDE.

-dsr-

[Peter Ehlert]

On 10/11/21 5:59 AM, local10 wrote:
> Hi,
>
> Am considering upgrading from buster to bullseye and looking for some feedback from those who have already done so. Was it worth it, in your opinion? Anything you like in particular about bullseye? Any problems/issues or degradation of performance/functionality/features, especially related to the KDE and its apps?

I am not a KDE user, all of my machines run Mate, so this is a bit off topic

Bullseye is great for me in almost every way.
The newer Mate apps are very nice, but not essential for me and my workflow.

next week I will deploy a "new" HP z820 work station that will be a
significant part of my LAN and private cloud. I don't know what GPU it
has yet, but it is Old Tech, 2012 vintage.
I am sure it will run on Bullseye just fine but...

However one of my favorite apps
(http://alarm-clock.pseudoberries.com/#intro) will not work.

this .deb works on Buster. but not Bullseye
https://packages.debian.org/stretch/amd64/alarm-clock-applet/download

so, for me, I believe I will go for Buster for a while.

Personal choice. A test install is advised.

[local10]

Oct 11, 2021, 14:21 by pb...@sdi-baja.com:

> A test install is advised.

That's my plan. Still interested to hear from more KDE users about their experiences though.

Thanks to everyone who responded.

[Anssi Saari]

local10 <loc...@tutanota.com> writes:

> Am considering upgrading from buster to bullseye and looking for some
> feedback from those who have already done so. Was it worth it, in your
> opinion?

Absolutely. I have a new video card in my desktop with no support in
Buster. Not even in backports. Could've hacked support in but didn't
want to and Bullseye release was close.

Everything else works pretty much as before. Minor tweak was needed due
to changes in newer org-mode. I actually cleaned up my system but didn't
quite get it out of FrankenDebian state. Have Megasync and Spotify
still. Also Debian Multimedia but I don't know if that's considered
FrankenDebian.

I don't run KDE in Debian, except for Konsole. The window manager on my
desktop is called Awesome and it hasn't been updated in a while so
obviously no change. Firefox, Thunderbird, mpv, also no change I
noticed. Well, I don't actually run Debian's packaged Firefox due to a
long standing issue. It's possible that has been fixed now so worth a
look.

[Махно]

I think the upgrade is worth it. Debian ąą is rock solid. I didn't notice practically any problems, except that libreoffice start up longer than in Debian 10.

[David Palacio]

Hi,

I run Bullseye and I find it very solid and trouble free. Though I didn't upgrade from Buster but instead used to run unstable and switched to stable when it released. Bullseye's KDE is very good and I have no issues with it other than I don't recommend you to run a Wayland session of it, you should stick with X11 for the time being. There exists unofficial repositories by a Debian/KDE packager if you wish to get more recent KDE software releases built for Bullseye.

[local10]

Oct 12, 2021, 13:40 by deb...@david.palacio.io:

> I run Bullseye and I find it very solid and trouble free. Though I didn't upgrade from Buster but instead used to run unstable and switched to stable when it released. Bullseye's KDE is very good and I have no issues with it other than I don't recommend you to run a Wayland session of it, you should stick with X11 for the time being. There exists unofficial repositories by a Debian/KDE packager if you wish to get more recent KDE software releases built for Bullseye.
>

Glad to hear that everything is well with KDE in bullseye, I'm running X11 too.