Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #238415 > unrolled thread

can't login via gdm

Back to article view | Back to linux.debian.user

Contents

  can't login via gdm Morgan Read <mstuff@read.org.nz> - 2021-08-10 09:40 +0200
    Re: can't login via gdm IL Ka <kazakevichilya@gmail.com> - 2021-08-10 12:40 +0200
      Re: can't login via gdm Zoltán Herman <zoltanhbz@gmail.com> - 2021-08-10 13:00 +0200
    Can't boot following re-install to LVM on LUKS [was: can't login via gdm] Morgan Read <mstuff@read.org.nz> - 2021-08-11 16:10 +0200
      Re: Can't boot following re-install to LVM on LUKS [was: can't login  via gdm] David Christensen <dpchrist@holgerdanske.com> - 2021-08-12 00:30 +0200
        Re: Can't boot following re-install to LVM on LUKS [was: can't login  via gdm] Morgan Read <morgan@read.nz> - 2021-08-16 16:50 +0200
          Re: Can't boot following re-install to LVM on LUKS [was: can't login  via gdm] "Andrew M.A. Cater" <amacater@einval.com> - 2021-08-16 17:20 +0200
            Re: Can't boot following re-install to LVM on LUKS [was: can't login  via gdm] Morgan Read <mstuff@read.org.nz> - 2021-08-19 12:30 +0200

#238415 — can't login via gdm

Hello List,

I ran the following to change my users UID & GIDs - but, now I have no
graphic gdm login.  I have no problem with a terminal login and the
same worked on a terminal based server.

After boot, gdm attempts to start but fails with a blinking cursor in
the top left corner - I can switch to tty2 etc.  Does anyone know what
stupidity I may have inflicted upon myself?

Commands executed:
# usermod -u 1003 guest
# usermod -u 1002 joe
# usermod -u 1001 blow
# groupmod -g 1003 guest
# groupmod -g 1002 joe
# groupmod -g 1001 blow
# find / -user 1002 -execdir chown 1003 '{}' \;
# find / -group 1002 -execdir chgrp 1003 '{}' \;
# find / -user 1001 -execdir chown 1002 '{}' \;
# find / -group 1001 -execdir chgrp 1002 '{}' \;
# find / -user 1000 -execdir chown 1001 '{}' \;
# find / -group 1000 -execdir chgrp 1001 '{}' \;

Thanks,
Morgan
-- 
Morgan Read
E mstuff@read.org.nz
(GM)

[toc] | [next] | [standalone]

#238420

[Multipart message — attachments visible in raw view] — view raw

>
>
> I ran the following to change my users UID & GIDs - but, now I have no
> graphic gdm login.  I have no problem with a terminal login and the
> same worked on a terminal based server.
>
>
Check your logs with journalctl.
There is also ~/.xsession-errors, but you probably lost access to your own
home dir.

As root do:
$ chown -R joe /home/joe
and login as joe

Create another user and try login with it. Does it work?

If it doesn't work:
Make sure that ``/tmp`` folder has full write permissions of everyone.
If not, fix it

Disable graphical login i.e. (``systemctl set-default multi-user.target``)
and reboot.
Try to start x with ``startx`` from the console.
Does it work? If not, check /var/log/Xorg.* log

[toc] | [prev] | [next] | [standalone]

#238421

[Multipart message — attachments visible in raw view] — view raw

or try login as root

IL Ka <kazakevichilya@gmail.com> ezt írta (időpont: 2021. aug. 10., K
12:35):

>
>> I ran the following to change my users UID & GIDs - but, now I have no
>> graphic gdm login.  I have no problem with a terminal login and the
>> same worked on a terminal based server.
>>
>>
> Check your logs with journalctl.
> There is also ~/.xsession-errors, but you probably lost access to your own
> home dir.
>
> As root do:
> $ chown -R joe /home/joe
> and login as joe
>
> Create another user and try login with it. Does it work?
>
> If it doesn't work:
> Make sure that ``/tmp`` folder has full write permissions of everyone.
> If not, fix it
>
> Disable graphical login i.e. (``systemctl set-default multi-user.target``)
> and reboot.
> Try to start x with ``startx`` from the console.
> Does it work? If not, check /var/log/Xorg.* log
>

[toc] | [prev] | [next] | [standalone]

#238523 — Can't boot following re-install to LVM on LUKS [was: can't login via gdm]

Hi List,
Since my cry for (fairly minor) help here:
https://lists.debian.org/debian-user/2021/08/msg00461.html
I think I've dug myself into a bit of a deep hole.

After having overcome a fairly fundamental bug with calamares as described here:
https://github.com/calamares/calamares/issues/1564#issuecomment-846321060
And, (unnecessarily as it turned out) re-installed my system, I find
I'm unable to boot.  Instead of getting a password prompt for my luks
lvm partition, I get dropped to the shell with the following error:

cryptsetup: ERROR: luks-[some UUID]: cryptsetup failed, bad password or options?
/bin/cat: /crypto_keyfile.bin: No such file or directory
Nothing to read on input

In
(initramfs) /cryptroot/crypttab
There is
luks-[some UUID] UUID=[some UUID] /crypto_keyfile.bin luks,keyscript=/bin/cat

I can manually mount my my luks volume so:
cryptsetup open /dev/sda3 --type luks [some UUID]
Enter passphrase for /dev/sda3:
To here:
/dev/mapper/[some UUID]
But, from there the LVM LVs don't get mounted and if I exit to boot
then I get dropped back to the shell.  And, lvm2 doesn't seem to be in
initramfs.

I can reboot to the live install usb image and when I do:
sudo cryptsetup open /dev/sda3 --type luks [some UUID]
Enter passphrase for /dev/sda3:
>From the terminal, then the lvm on luks volumes pop up as expected
immediately decrypted, named under /dev/mapped/[some VG-LV-name],
/dev/[some VG]/[some LV], enumerated at /dev/dm-xyz etc and mounted
under /media/user/[partition label]

So, may be calamares has more than one bug with LVM on LUKS?  Or, as
the above bug only refers to LVM and not LUKS, perhaps I've done (or
not done) something silly during the install with calamares and LUKS
(though, it does seem a pretty pedestrian walk through of the install
process).

I guess, there's something not referenced in the initramfs?

Any pointers to a way back from Armageddon would be great :)

Thanks.
--
Morgan Read
E mstuff@read.org.nz
(GM)

[toc] | [prev] | [next] | [standalone]

#238546 — Re: Can't boot following re-install to LVM on LUKS [was: can't login via gdm]

On 8/11/21 6:45 AM, Morgan Read wrote:
> Hi List,
> Since my cry for (fairly minor) help here:
> https://lists.debian.org/debian-user/2021/08/msg00461.html
> I think I've dug myself into a bit of a deep hole.
> 
> After having overcome a fairly fundamental bug with calamares as described here:
> https://github.com/calamares/calamares/issues/1564#issuecomment-846321060
> And, (unnecessarily as it turned out) re-installed my system, I find
> I'm unable to boot. ...


"Find the needle in the haystack" does not work for me:

1.  It takes an unpredictable amount of time.

2.  Meanwhile, operations are stopped.

3.  Only one needle?  Really?

4.  How do I validate that the haystack is "correct"?


I keep my operating system images small enough to fit onto a single "16 
GB" device (USB flash drive, SD card, SSD, HDD, etc.) using BIOS, MBR, 
and no hardware or software RAID (lowest common denominator).  I take 
raw binary images of my operating system devices periodically and as 
needed.  I keep my operating system configuration files in a version 
control system.  I keep the vast majority of my data on RAID in a file 
server.  I snapshot, replicate, back up, archive, etc., everything and 
rotate backup media on-, near-, and off-site.  I have redundant and 
spare hardware.


When a system image is damaged or doubtful, I restore the last raw 
binary image onto a blank device, check out the configuration files, and 
restore local data.  The computer is back in operation in a predictable 
amount of time with a high level of confidence that everything is correct.


David

[toc] | [prev] | [next] | [standalone]

#238827 — Re: Can't boot following re-install to LVM on LUKS [was: can't login via gdm]

[Multipart message — attachments visible in raw view] — view raw

On 11/08/2021 11:30 pm, David Christensen wrote:
> On 8/11/21 6:45 AM, Morgan Read wrote:
>> After having overcome a fairly fundamental bug with calamares as
>> described here:
>> https://github.com/calamares/calamares/issues/1564#issuecomment-846321060
>> And, (unnecessarily as it turned out) re-installed my system, I find
>> I'm unable to boot. ...
...
> When a system image is damaged or doubtful, I restore the last raw
> binary image onto a blank device, check out the configuration files, and
> restore local data. The computer is back in operation in a predictable
> amount of time with a high level of confidence that everything is correct.

Thanks David - I do have my home directories all comfortably backed-up
to a raid device on my local server - I stopped taking stuff off-site a
year or so ago. However, it's still a right royal pita when both the
live CD and full install DVD are broken, see commentary from here:
https://github.com/calamares/calamares/issues/1564#issuecomment-898246354

I do find it puzzling that there's not straight forward way of doing
LVM-on-LUKS.

Regards,
-- 
Morgan Read

[toc] | [prev] | [next] | [standalone]

#238828 — Re: Can't boot following re-install to LVM on LUKS [was: can't login via gdm]

On Mon, Aug 16, 2021 at 03:46:51PM +0100, Morgan Read wrote:
> On 11/08/2021 11:30 pm, David Christensen wrote:
> > On 8/11/21 6:45 AM, Morgan Read wrote:
> >> After having overcome a fairly fundamental bug with calamares as
> >> described here:
> >> https://github.com/calamares/calamares/issues/1564#issuecomment-846321060
> >> And, (unnecessarily as it turned out) re-installed my system, I find
> >> I'm unable to boot. ...
> ...

Morgan,

Possibly just don't use calamares. Use the standard Debian installer (d-i for
short). Either the standard or expert installs offer you "ordinary" LVM or
encrypted LVM with LUKS, made even more straightforward if you can choose
guided partitioning. Do be aware, however, that if you choose 
multi-filesystem, then you might have to check that the appropriate 
partition sizes are correct for you. In testing, we often have to swap the 
mount point labels around.

Also: the standard size for swap is now 1G and for a dedicated /tmp 
partition of 2G.

All the very best, as ever,

Andy Cater


> > When a system image is damaged or doubtful, I restore the last raw
> > binary image onto a blank device, check out the configuration files, and
> > restore local data. The computer is back in operation in a predictable
> > amount of time with a high level of confidence that everything is correct.
> 
> Thanks David - I do have my home directories all comfortably backed-up
> to a raid device on my local server - I stopped taking stuff off-site a
> year or so ago. However, it's still a right royal pita when both the
> live CD and full install DVD are broken, see commentary from here:
> https://github.com/calamares/calamares/issues/1564#issuecomment-898246354
> 
> I do find it puzzling that there's not straight forward way of doing
> LVM-on-LUKS.
> 
> Regards,
> -- 
> Morgan Read
> 

[toc] | [prev] | [next] | [standalone]

#239028 — Re: Can't boot following re-install to LVM on LUKS [was: can't login via gdm]

[Converted from HTML] — view raw

> On Mon, Aug 16, 2021 at 03:46:51PM +0100, Morgan Read wrote:
> > On 11/08/2021 11:30 pm, David Christensen wrote:
> > > On 8/11/21 6:45 AM, Morgan Read wrote:
> > >> After having overcome a fairly fundamental bug with calamares as
> > >> described here:
> > >> [https://github.com/calamares/calamares/issues/1564#issuecomment-846321
> 060][1]
> > >> And, (unnecessarily as it turned out) re-installed my system, I find
> > >> I'm unable to boot. ...
> > ...
> 
> Morgan,

Hi Andy, Thanks for coming back to me:-

> Possibly just don't use calamares. Use the standard Debian installer (d-i fo
> r
> short). Either the standard or expert installs offer you "ordinary" LVM or
> encrypted LVM with LUKS, made even more straightforward if you can choose

Actually, it was calamares that worked better than d-i - after failing with
Debian's calamares live install I followed the instructions for d-i here:
[https://www.blakehartshorn.com/installing-debian-on-existing-encrypted-lvm/][
2]
Blake Hartshorn seems to do a pretty good job of describing the problem and
solution.  However, it wasn't a solution for me because the install failed
repeatedly over the course of a couple of days (I was using 10.10 cf. BH using
10.4) - it wasn't until after I'd invested in a new flash drive and continued
with the same fails that I tried the PureOS live install, which worked.  You
can read a full discussion of my woes here:
[https://github.com/calamares/calamares/issues/1564#issuecomment-898246354][3]
Essentially, d-i couldn't get passed either the base system dpkg install (most
usually base-passwd) or the following software install.  The dpkg warnings
were thrown on "parsing file '/var/lib/dpkg/status' near line 5 package 'dpkg'
: missing description field" with errors on the stateoverride file.

I would have filed bug reports but then Debian went to 11 and I figured
there'd be a new disk spun so it was all history anyway and I had a working
machine with PureOS.


> Also: the standard size for swap is now 1G and for a dedicated /tmp 
> partition of 2G.

That's interesting - so nothing for hibernation on swap?  I gave 8G to allow
for hibernation, but would happily give it away if I could?

Many thanks.

-- 
Morgan Read

UNITED KINGDOM
Em: <email:mstuffATreadDOTorgDOTnz>

Confused about DRM?
Get all the info you need at:
[<http://drm.info/>][4]

[1]: https://github.com/calamares/calamares/issues/1564#issuecomment-846321060
[2]: https://www.blakehartshorn.com/installing-debian-on-existing-encrypted-lv
m/
[3]: https://github.com/calamares/calamares/issues/1564#issuecomment-898246354
[4]: http://drm.info/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web