Crap, I upgraded too and now it doesn’t boot anymore (Librem 15 v4). No login screen comes up. Only a black screen with the mouse cursor. I can go into the console with Ctrl-Alt-F2 but that’s all. No graphical interface anymore.
Doing a /etc/init.d/gdm3 stop
and /etc/init.d/gdm3 start
gives a “Something went wrong, contact a system administrator” screen. I’d rather not try to mix in Debian packages, as I don’t want to risk ending up in an even more inconsistent state. But hope this will be solved soon
What did you upgrade? gnome-shell from Debian testing? how did you do this?
The way I did it was:
sudo su
echo "deb http://deb.debian.org/debian/ testing main" > /etc/apt/sources.list.d/debiantesting.list
apt update
apt -t testing install gnome-shell
rm /etc/apt/sources.list.d/debiantesting.list
apt update
I did from unstable really because I’m using Debian testing and when I did it it wasn’t fixed yet.
PS: This is just a workaround do it under your own resposability.
I don’t know if @jeremiah is aware about this
If your system is working don’t update it until this is fixed.
You could try to debug it using the systemd debug shell after disabling the messages written to it.
@ChriChri I managed to solve my problem described above without the need for this - But your hint is greatly appreciated, I am sure it might be useful for me later. Thank you!
In case somebody has the same problem as I described above, i.e., that you cannot log in to the shell after switching away from X by pressing Ctrl + Alt + F2
. My password can apparently not be entered successfully because I’m somehow using an unsupported string of special characters. This by itself is wild already…
Anyway, I ended up working around this issue by booting into the recovery shell. From there I created a new user by running
adduser username
Set a simple® password, then add this user to the sudo group by running
usermod -aG sudo username
Now I can at least log in with this new user on the shell and do everything, but still no fix to X not loading. Any update? Any other solutions than switching over to debian testing? Also, could this easily be undone once the PureOS repos are fixed / the problem is solved?
not to sound out of touch, is everyone trying this because pureos amber is not being updated, or that byzantium will take a long time to fully integrate? is it recommended to try this or to wait for offiicial pureos to move to byzantium?
PureOS byzantium is based on Debian testing that comes with newest packages (and problems) the best is to stay in PureOS amber if you don’t need/want a ‘cutting edge’ system and you don’t want to deal with a broken system.
PureOS Amber will be always Amber.
thank you for the reply. so at some point down the road, those running pureos amber will be able to upgrade to a (stable) pureos byzantium?
Yes and no, Amber is Amber always but the packages in Byzantium will be in Amber when they are enought stable to avoid this kind of errors
This is not quite accurate example. Packages would not go from B to A automatically, only minor version/patchlevel upgrades. Major version bump wont go, they will rather to to the next stable release (if any). For those packages which bumped to next major there will only be backports of necessary features and security mitigations. But again feature backport most likely only if it applies to older codebase, no one will be spending hours to re-implement the feature if upstream codebase diverges significantly.
so when byzantium becomes stable, a full upgrade from amber is the recommended step? (for those who require high stability)
Byzantium will become the new stable at some point, and there will be a new rolling release
But definitely not yet.
Byzantium is not supposed to become stable, as rolling release it will be constantly receiving new packages remaining on the cutting edge. Like Sid. But presumably more stable.
Sid is always sid, sid periodically is snapped into new release which becomes new testing which then stabilize and becomes new stable. So there was buster testing which became buster stable and bullseye testing was snapped, bullseye becomes stable and bookworm is snapped (currently there’s only name).
In PureOS stable is based on debian stable but Byzantium is Purism’s own spin, like Sid-after-certain-testing.
Uff, ok, so I’m wrong then and purism lied about rolling release Rolling cannot become stable, by definition. Sid cannot become testing.
ok thank you, so to be clear, staying on amber is recommended for stability, and byzantium for bleeding edge? So at some point in the future, will there be a stable release based on byzantium, or will the next pureos be based on debian stable? I apologize for the questions, but its pretty interesting and slightly confusing.
This discussion could become confusing because Purism has not yet said what the naming convention will be when the opportunity is there to move to a new stable release, at least as far as I know.
In principle, Amber could remain forever stable and Byzantium could remain forever rolling - even though both jump to new releases.
Or, as hinted at with the naming, and probably safer and more consistent with other distro practice, Purism could move to C* and D* colors - so that Byzantium remains a rolling release but is in fact relatively stable and dormant (perhaps only receiving required security updates for a period of time) and all Amber users should upgrade to C* while all Byzantium users should upgrade to D*.
Then the distro itself will be release-based but the name will be rolling (alias). It’s the same as if you call debian-stable and debian-testing to be rolling releases but they are not, they are merely pointers to current release snapshot. It is stabilisation (freeze) phase which differs true rolling from release-based distribution. Freeze means release stops receiving new packages, only patches, thus diverging further and further from the upstream.
I was trying to use once pointing to the alias rather than specific release, but in the essence it doesn’t solve anything - you still have dist-upgrade hassle on each release cycle.
I did a fresh install of PureOS Amber in order to have a working system. Now I’m wondering how I can know when I can upgrade to Byzantium without bricking my system again? Does this dependency problem effect everyone on Byzantium doing the update? It seems not many people are effected by this.
I probably haven’t been affected thanks to reports from Linuxnoob and you, so I stopped applying updates daily, as my Librem with PureOS is the only machine I can currently use for working from home.
I looked again at Byzantium updates for my system a couple of hours ago, but I haven’t been able to infer from version numbers changes and Debian error report mentioned in Gnomeshell crashes after update from 25/26th of april - byzantium whether it has been fixed in Byzantium.