Hi here,

For Linux desktop login, say, Gnome, with the bash as the default shell, is there a method to save all the detailed sourced files information and the errors during the logging process?

I often encounter strange phenomena and problems in login, but I don't know where they come from. Any tips will be appreciated.

Regards,
Zhao

On Wed, 25 Jan 2023 20:28:31 -0800 (PST), hongy...@gmail.com wrote:
> Hi here,
>
> For Linux desktop login, say, Gnome, with the bash as the default shell, is there a method to save all the detailed sourced files information and the errors during the logging process?
>
> I often encounter strange phenomena and problems in login, but I don't know where they come from. Any tips will be appreciated.

Since you did not provide what Distribution you are using, no one can a
lot of detailed information.

So, check system logs, X11 logs, and user logs.
Since I am running Xfce4 on Mageia Linux that would be journalctl, /var/log.
and ~/.*error*

On Thursday, January 26, 2023 at 1:13:16 PM UTC+8, Bit Twister wrote:
> On Wed, 25 Jan 2023 20:28:31 -0800 (PST), hongy...@gmail.com wrote:
> > Hi here,
> >
> > For Linux desktop login, say, Gnome, with the bash as the default shell, is there a method to save all the detailed sourced files information and the errors during the logging process?
> >
> > I often encounter strange phenomena and problems in login, but I don't know where they come from. Any tips will be appreciated.
> Since you did not provide what Distribution you are using, no one can a
> lot of detailed information.

No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 22.10
Release: 22.10
Codename: kinetic

werner@X10DAi:~$ echo $SHELL
/bin/bash

Furthermore, I've complicated and structured customized profile configurations in the following locations:

werner@X10DAi:~$ ls /etc/profile.d/000-rootshare.sh
/etc/profile.d/000-rootshare.sh
werner@X10DAi:~$ ls ~/.profile.d/
000-networkd-resolved-netplan-dnsmasq.bash 300-fcitx5-rime.bash
001-dnsmasq-dnsproxy.bash 300-firefox.bash
005-blacklist-nouveau.sh 300-ftl.git.bash
005-disable-update-upgrade.sh 300-fzf.git.bash
005-fix-locale.bash 300-gnus.d.bash
005-git-env.sh 300-iso.bash
005-zsys-daemon-connection-error.bash 300-nginx.git.bash
010-docker-client.sh 300-openjdk.bash
010-docker-daemon.sh 300-openssh.bash
010-dockerd-data-root.sh.save 300-ripgrep-all.git.bash.save
010-proxychains-ng-daemon.git.bash 300-ripgrep.git.bash
015-apollo.git.bash 300-slurm.bash
100-gnome-terminal.sh 300-tclreadline.git.bash
100-xdgbds.bash 300-tessdata_best.git.bash.save
200-git-config.sh 300-tessdata.git.bash
200-go.git.bash 300-texlive.bash
200-pyenv.git.bash 300-wine-libgl.bash
300-brook.git.bash 300-x11docker.git.bash
300-calibre.git.bash 400-maven.bash
300-carat.git.bash 900-homeshare.git.bash
300-ext-saladict-chrome.bash

> So, check system logs, X11 logs, and user logs.
> Since I am running Xfce4 on Mageia Linux that would be journalctl, /var/log.
> and ~/.*error*

Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.

[1] https://askubuntu.com/questions/587868/are-you-sure-you-want-to-proceed-message-on-login

Regards,
Zhao

On Wed, 25 Jan 2023 23:17:06 -0800 (PST), "hongy...@gmail.com"
<hongyi.zhao@gmail.com> wrote:

[...]
>werner@X10DAi:~$ ls /etc/profile.d/000-rootshare.sh
>/etc/profile.d/000-rootshare.sh
>werner@X10DAi:~$ ls ~/.profile.d/
>000-networkd-resolved-netplan-dnsmasq.bash 300-fcitx5-rime.bash
>001-dnsmasq-dnsproxy.bash 300-firefox.bash
[...]

Binay search will take just a few tries.
Move half of the scripts in ~/.profile.d/ somewhere where they
will not run.
Try again.
- No error: the problem was in the half you moved away
- Still error: the problem was in the half you did not move

Proceed by activating half of the set of scripts that has the
problem.
Repeat until the size of the set is one script.
--
Kees Nuyt

On Thursday, January 26, 2023 at 8:12:15 PM UTC+8, Kees Nuyt wrote:
> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), "hongy...@gmail.com"
> <hongy...@gmail.com> wrote:
>
> [...]
> >werner@X10DAi:~$ ls /etc/profile.d/000-rootshare.sh
> >/etc/profile.d/000-rootshare.sh
> >werner@X10DAi:~$ ls ~/.profile.d/
> >000-networkd-resolved-netplan-dnsmasq.bash 300-fcitx5-rime.bash
> >001-dnsmasq-dnsproxy.bash 300-firefox.bash
> [...]
>
> Binay search will take just a few tries.
> Move half of the scripts in ~/.profile.d/ somewhere where they
> will not run.
> Try again.
> - No error: the problem was in the half you moved away
> - Still error: the problem was in the half you did not move
>
> Proceed by activating half of the set of scripts that has the
> problem.
> Repeat until the size of the set is one script.

Nice idea. I'll give a try.

> --
> Kees Nuyt

Regards,
Zhao

On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:

> Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.

I suggest creating a test account, say junk. Log out of hongyi and into junk.
If junk has the same problem as hongyi you know it is a system wide problem.
If junk does not have the problem then you know it is a user (hongyi) problem.

You can test the login process from either account by logging into the
other user account by clicking up a terminal and doing
su - junk
from a hongyi terminal or su - hongyi from a junk terminal, depending
on who you want to test.

For you, I suggest anytime you make any login script change is to
su - junk and very there are no problems.

On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>
> > Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.
> I suggest creating a test account, say junk. Log out of hongyi and into junk.
> If junk has the same problem as hongyi you know it is a system wide problem.
> If junk does not have the problem then you know it is a user (hongyi) problem.
>
> You can test the login process from either account by logging into the
> other user account by clicking up a terminal and doing
> su - junk
> from a hongyi terminal or su - hongyi from a junk terminal, depending
> on who you want to test.
>
> For you, I suggest anytime you make any login script change is to
> su - junk and very there are no problems.

Do you mean always first test any newly added/revised script with this account?

Best,
Zhao

On Thu, 26 Jan 2023 19:09:46 -0800 (PST), hongy...@gmail.com wrote:
> On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
>> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:

>> For you, I suggest anytime you make any login script change is to
>> su - junk and very there are no problems.
>
> Do you mean always first test any newly added/revised script with this
> account?

Why not. Can not see that it take long to su -l junk and enter password.

On my distribution, that kind of error will not let you do a gui login.

What is your procedure if you make a change and can not do a gui login?

On 27.01.2023 04:09, hongy...@gmail.com wrote:
> On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
>> [...]
>
> Do you mean always first test any newly added/revised script with this account?

Nah! In this part of the world we say that testing is only for wimps.

We accept that systems get spoiled. Provide arbitrary changes quickly.
It's certainly thrilling.

Unavailable and unreliable and non-trustworthy environments for users
and customers are just a minor nuisance.

Real heroes can then jump in an fix any issues. (With more untested
scripts.)

Mind that admins get no credit for nicely and reliably working systems.

If you show that you can fix issues (and issues of fixes, recursively)
you prove that you are a competent doer, a hero with quick solutions.
(In many environments no one cares if you are the [continuous] source
of the issue.)

This all works well at least in companies with ignorant management and
where folks have to depend on skills and expertise.

Disclaimer: Anyone who is taking above as serious professional approach
is hopelessly lost. - OTOH that might fit well in quite some parts of
our (IT-)world. *sigh*

Janis

>
> Best,
> Zhao
>

On Friday, January 27, 2023 at 11:16:01 AM UTC+8, Bit Twister wrote:
> On Thu, 26 Jan 2023 19:09:46 -0800 (PST), hongy...@gmail.com wrote:
> > On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> >> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>
> >> For you, I suggest anytime you make any login script change is to
> >> su - junk and very there are no problems.
> >
> > Do you mean always first test any newly added/revised script with this
> > account?
> Why not. Can not see that it take long to su -l junk and enter password.

Agreed.
> On my distribution, that kind of error will not let you do a gui login.

That's the same dilemma I face here.

> What is your procedure if you make a change and can not do a gui login?

Fortunately, I found the following quick workaround to avoid rebooting the system for the next login attempt when the "Are you sure you want to proceed?" problem occurs:

1. Ctrl + Alt + F4 opens a console window, where you can log in using the username and password.
2. After logging in, run the following command to restart gdm:

$ sudo systemctl restart gdm

3. Then you will have the opportunity to log back in to the GUI.

Best,
Zhao

On Fri, 27 Jan 2023 00:35:24 -0800 (PST), hongy...@gmail.com wrote:
> On Friday, January 27, 2023 at 11:16:01 AM UTC+8, Bit Twister wrote:
>> On Thu, 26 Jan 2023 19:09:46 -0800 (PST), hongy...@gmail.com wrote:
>> > On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
>> >> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>>
>> >> For you, I suggest anytime you make any login script change is to
>> >> su - junk and very there are no problems.
>> >
>> > Do you mean always first test any newly added/revised script with this
>> > account?
>> Why not. Can not see that it take long to su -l junk and enter password.
>
> Agreed.
>
>> On my distribution, that kind of error will not let you do a gui login.
>
> That's the same dilemma I face here.
>
>> What is your procedure if you make a change and can not do a gui login?
>
> Fortunately, I found the following quick workaround to avoid rebooting the system for the next login attempt when the "Are you sure you want to proceed?" problem occurs:
>
> 1. Ctrl + Alt + F4 opens a console window, where you can log in using the username and password.
> 2. After logging in, run the following command to restart gdm:
>
> $ sudo systemctl restart gdm
>
> 3. Then you will have the opportunity to log back in to the GUI.

But what happens if gui does not start.

Can YOU run a non gui editor to rectify your induced problem?

On Friday, January 27, 2023 at 6:50:08 PM UTC+8, Bit Twister wrote:
> On Fri, 27 Jan 2023 00:35:24 -0800 (PST), hongy...@gmail.com wrote:
> > On Friday, January 27, 2023 at 11:16:01 AM UTC+8, Bit Twister wrote:
> >> On Thu, 26 Jan 2023 19:09:46 -0800 (PST), hongy...@gmail.com wrote:
> >> > On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> >> >> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
> >>
> >> >> For you, I suggest anytime you make any login script change is to
> >> >> su - junk and very there are no problems.
> >> >
> >> > Do you mean always first test any newly added/revised script with this
> >> > account?
> >> Why not. Can not see that it take long to su -l junk and enter password.
> >
> > Agreed.
> >
> >> On my distribution, that kind of error will not let you do a gui login.
> >
> > That's the same dilemma I face here.
> >
> >> What is your procedure if you make a change and can not do a gui login?
> >
> > Fortunately, I found the following quick workaround to avoid rebooting the system for the next login attempt when the "Are you sure you want to proceed?" problem occurs:
> >
> > 1. Ctrl + Alt + F4 opens a console window, where you can log in using the username and password.
> > 2. After logging in, run the following command to restart gdm:
> >
> > $ sudo systemctl restart gdm
> >
> > 3. Then you will have the opportunity to log back in to the GUI.
> But what happens if gui does not start.

As far as the current focus question is concerned, a.k.a., "Are you sure you want to proceed?", I really didn't have observed GUI did not start again.
> Can YOU run a non gui editor to rectify your induced problem?

Only if I know what's the culprit, otherwise, I don't know what should be rectified, just as the question discussed here.

Regards,
Zhao

On Fri, 27 Jan 2023 06:14:53 -0800 (PST), hongy...@gmail.com wrote:
> On Friday, January 27, 2023 at 6:50:08 PM UTC+8, Bit Twister wrote:

>> Can YOU run a non gui editor to rectify your induced problem?
>
> Only if I know what's the culprit, otherwise, I don't know what should be rectified, just as the question discussed here.

If you are modifying one of your files in the login path, I assume you have
set -u
just after the #!/bin/bash at the top of your scripts, I would expect when
you do a su - l junk you will see some bash error message.

The quest was not about what file, but can YOU run a non-gui editor?

On Saturday, January 28, 2023 at 2:44:59 AM UTC+8, Bit Twister wrote:
> On Fri, 27 Jan 2023 06:14:53 -0800 (PST), hongy...@gmail.com wrote:
> > On Friday, January 27, 2023 at 6:50:08 PM UTC+8, Bit Twister wrote:
>
> >> Can YOU run a non gui editor to rectify your induced problem?
> >
> > Only if I know what's the culprit, otherwise, I don't know what should be rectified, just as the question discussed here.
> If you are modifying one of your files in the login path, I assume you have
> set -u
> just after the #!/bin/bash at the top of your scripts, I would expect when
> you do a su - l junk you will see some bash error message.

Thank you for this tip.
> The quest was not about what file, but can YOU run a non-gui editor?

Yes. I can.

Bit Twister <BitTwister@mouse-potato.com> wrote:
> On Thu, 26 Jan 2023 19:09:46 -0800 (PST), hongy...@gmail.com wrote:
>> On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
>>> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>
>>> For you, I suggest anytime you make any login script change is to
>>> su - junk and very there are no problems.
>>
>> Do you mean always first test any newly added/revised script with this
>> account?
>
> Why not. Can not see that it take long to su -l junk and enter password.
>
> On my distribution, that kind of error will not let you do a gui login.
>
>
> What is your procedure if you make a change and can not do a gui login?

I run a second X server, when I make changes I test by switching to it
& logging in. Most likely, if login fails, it's the change I just made.
I also use it to test changes to my FVWM configuration to see if I
like them.

Jerry

On Thursday, January 26, 2023 at 8:12:15 PM UTC+8, Kees Nuyt wrote:
> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), "hongy...@gmail.com"
> <hongy...@gmail.com> wrote:
>
> [...]
> >werner@X10DAi:~$ ls /etc/profile.d/000-rootshare.sh
> >/etc/profile.d/000-rootshare.sh
> >werner@X10DAi:~$ ls ~/.profile.d/
> >000-networkd-resolved-netplan-dnsmasq.bash 300-fcitx5-rime.bash
> >001-dnsmasq-dnsproxy.bash 300-firefox.bash
> [...]
>
> Binay search will take just a few tries.
> Move half of the scripts in ~/.profile.d/ somewhere where they
> will not run.
> Try again.
> - No error: the problem was in the half you moved away
> - Still error: the problem was in the half you did not move
>
> Proceed by activating half of the set of scripts that has the
> problem.
> Repeat until the size of the set is one script.

I've tried out the above method, and find that the "Are you sure you want to proceed?" problem is not caused by any scripts located under ~/.profile.d/. So what should I do for further examination?

> --
> Kees Nuyt
Zhao

On Sun, 29 Jan 2023 17:34:18 -0800 (PST), "hongy...@gmail.com"
<hongyi.zhao@gmail.com> wrote:

> I've tried out the above method, and find that the
> "Are you sure you want to proceed?" problem is not
> caused by any scripts located under ~/.profile.d/. So
> what should I do for further examination?

Continue with the scripts in /etc/profile.d/ , and
and ~/.xxxxxx scripts, like .bashrc and .profile,
and any GUI equivalents of those.
--
Kees Nuyt

On Monday, January 30, 2023 at 5:42:13 PM UTC+8, Kees Nuyt wrote:
> On Sun, 29 Jan 2023 17:34:18 -0800 (PST), "hongy...@gmail.com"
> <hongy...@gmail.com> wrote:
>
> > I've tried out the above method, and find that the
> > "Are you sure you want to proceed?" problem is not
> > caused by any scripts located under ~/.profile.d/. So
> > what should I do for further examination?
> Continue with the scripts in /etc/profile.d/ , and
> and ~/.xxxxxx scripts, like .bashrc and .profile,
> and any GUI equivalents of those.

In my case, it seems that all stuff under /etc/profile.d/, and in .bashrc or .profile are not the culprits. And I've the following ~/.xxxxxx scripts:

werner@X10DAi:~$ find -L . -maxdepth 1 -mindepth 1 -xtype l -name '.*'
../.bashrc
../.newsrc
../.newsrc.eld
../.gprc
../.tclshrc
../.gnus.el
../.git
../.zshrc
werner@X10DAi:~$ find . -maxdepth 1 -mindepth 1 -type f -name '.*'
../.sys1og.conf
../.pypirc
../.xdvirc
../.zlua
../.sudo_as_admin_successful
../.bash_history
../.wget-hsts
../.pmgrc.yaml
../.latexmkrc
../.profile
../.gap_hist
../.gp_history
../.ripgreprc
../.emacs-w3m.el
../.condarc
../.lesshst
../.python_history
../.gitconfig
../.xinputrc

However, not all of them will be sourced by the login process. So how should I further focus on some of them?

> --
> Kees Nuyt
Zhao

On Monday, January 30, 2023 at 9:34:21 AM UTC+8, hongy...@gmail.com wrote:
> On Thursday, January 26, 2023 at 8:12:15 PM UTC+8, Kees Nuyt wrote:
> > On Wed, 25 Jan 2023 23:17:06 -0800 (PST), "hongy...@gmail.com"
> > <hongy...@gmail.com> wrote:
> >
> > [...]
> > >werner@X10DAi:~$ ls /etc/profile.d/000-rootshare.sh
> > >/etc/profile.d/000-rootshare.sh
> > >werner@X10DAi:~$ ls ~/.profile.d/
> > >000-networkd-resolved-netplan-dnsmasq.bash 300-fcitx5-rime.bash
> > >001-dnsmasq-dnsproxy.bash 300-firefox.bash
> > [...]
> >
> > Binay search will take just a few tries.
> > Move half of the scripts in ~/.profile.d/ somewhere where they
> > will not run.
> > Try again.
> > - No error: the problem was in the half you moved away
> > - Still error: the problem was in the half you did not move
> >
> > Proceed by activating half of the set of scripts that has the
> > problem.
> > Repeat until the size of the set is one script.
> I've tried out the above method, and find that the "Are you sure you want to proceed?" problem is not caused by any scripts located under ~/.profile.d/. So what should I do for further examination?

Finally, it turns out that the problem is caused by a bug of a Ubuntu patch and the incorrect settings in my initialization scripts. See the following links for more details:

https://git.launchpad.net/ubuntu/+source/gdm3/tree/debian/patches/ubuntu/config_error_dialog.patch
https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/2005135
https://discourse.gnome.org/t/are-you-sure-you-want-to-proceed-message-on-login/13758/7

> > --
> > Kees Nuyt
> Zhao

On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>
> > Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.
> I suggest creating a test account, say junk. Log out of hongyi and into junk.
> If junk has the same problem as hongyi you know it is a system wide problem.
> If junk does not have the problem then you know it is a user (hongyi) problem.
>
> You can test the login process from either account by logging into the
> other user account by clicking up a terminal and doing
> su - junk
> from a hongyi terminal or su - hongyi from a junk terminal, depending
> on who you want to test.
>
> For you, I suggest anytime you make any login script change is to
> su - junk and very there are no problems.

According to the above suggestions, I conducted the following test:

werner@X10DAi:~$ sudo useradd test
werner@X10DAi:~$ sudo passwd test # Set the pw as: asdf1234
werner@X10DAi:~$ sudo usermod -d /home/test -s /bin/bash test
werner@X10DAi:~$ sudo mkdir /home/test

# This time, the problem discussed here doesn't appear again:
werner@X10DAi:~$ su - test
Password:
test@X10DAi:~$

So, I tried to login back as follows and the problem appeared once more:

test@X10DAi:~$ su - werner
Password:
terminate called after throwing an instance of 'std::runtime_error'
what(): Failed to create dbus connection
xhost: unable to open display ""
dirname: invalid option -- 'b'
Try 'dirname --help' for more information.
readlink: invalid option -- 'b'
Try 'readlink --help' for more information.

Any tips for further debugging?

Regards,
Zhao

On Sun, 5 Feb 2023 03:41:22 -0800 (PST), "hongy...@gmail.com"
<hongyi.zhao@gmail.com> wrote:

> Finally, it turns out that the problem is caused by a bug
> of a Ubuntu patch and the incorrect settings in my
> initialization scripts.

I'm glad you could solve it.
--
Kees Nuyt

On Monday, February 6, 2023 at 10:15:00 PM UTC+8, Kees Nuyt wrote:
> On Sun, 5 Feb 2023 03:41:22 -0800 (PST), "hongy...@gmail.com"
> <hongy...@gmail.com> wrote:
>
> > Finally, it turns out that the problem is caused by a bug
> > of a Ubuntu patch and the incorrect settings in my
> > initialization scripts.
> I'm glad you could solve it.

But Gunnar Hjalmarsson <2005135@bugs.launchpad.net> gives the alternative maybe nicer solution *:

--- begin quote ---
Gunnar Hjalmarsson (gunnarhj) wrote on 2020-07-12: #2

Second thoughts...

The --no-markup option prevents zenity from interpreting \n which makes the dialog hard to read in most cases. So only adding --no-markup is not a viable solution.

As a possible alternative I played with sed() a bit. The script includes this line:

TEXT="$PARA1\n\n$(fold -s $ERR)\n\n$PARA2"

My thought is to add a new line right after that. Something along these lines:

TEXT=$(printf '%s' "$TEXT" | sed 's/</\&lt;/g')

which would replace all occurrences of '<' with '&lt;'. Can you please test that and see if it helps in your case.
Changed in gdm3 (Ubuntu):
status: In Progress → Incomplete
Changed in lightdm (Ubuntu):
status: In Progress → Incomplete
summary: - lightdm config-error-dialog.sh should pass --no-markup
+ config-error-dialog.sh should pass --no-markup
--- end quote ---

Below is my comment on the above method:

--- begin quote ---
Thank you for letting me know your alternative nice solution. But
there are so many HTML Character Entities [1-2]. How does your method
deal with all these possible situations?

[1] https://www.w3schools.com/HTML/html_entities.asp
[2] https://stackoverflow.com/questions/5068951/what-do-lt-and-gt-stand-for
--- end quote ---

Any more comments, suggestions, and alternative enhancements for fixing this bug gracefully and thoroughly will be greatly appreciated.

* https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1887252
> --
> Kees Nuyt

Regards,
Zhao

On Mon, 6 Feb 2023 01:40:42 -0800 (PST), hongy...@gmail.com wrote:
> On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
>> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
>>
>> > Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.
>> I suggest creating a test account, say junk. Log out of hongyi and into junk.
>> If junk has the same problem as hongyi you know it is a system wide problem.
>> If junk does not have the problem then you know it is a user (hongyi) problem.
>>
>> You can test the login process from either account by logging into the
>> other user account by clicking up a terminal and doing
>> su - junk
>> from a hongyi terminal or su - hongyi from a junk terminal, depending
>> on who you want to test.
>>
>> For you, I suggest anytime you make any login script change is to
>> su - junk and very there are no problems.
>
> According to the above suggestions, I conducted the following test:
>
> werner@X10DAi:~$ sudo useradd test
> werner@X10DAi:~$ sudo passwd test # Set the pw as: asdf1234
> werner@X10DAi:~$ sudo usermod -d /home/test -s /bin/bash test
> werner@X10DAi:~$ sudo mkdir /home/test
>
> # This time, the problem discussed here doesn't appear again:
> werner@X10DAi:~$ su - test
> Password:
> test@X10DAi:~$
>
> So, I tried to login back as follows and the problem appeared once more:
>
> test@X10DAi:~$ su - werner
> Password:
> terminate called after throwing an instance of 'std::runtime_error'
> what(): Failed to create dbus connection
> xhost: unable to open display ""
> dirname: invalid option -- 'b'
> Try 'dirname --help' for more information.
> readlink: invalid option -- 'b'
> Try 'readlink --help' for more information.
>su - werner
> Any tips for further debugging?

if su - test works and su - werner failed
then some script executed by werner login is failing.

No idea why

Message about dirname and readlink aught to be fixed.

Apparently you are running some script executing those instructions that
test does not execute.

By the way, after spending years as an maintenance programmer I found it beneficial
to fix any errors I find even if I believe they have no bearing on the problem
I am trying to fix.

For your xhost problem I would need to see the xhost command. Looking through my
host xx__local.sh script I have
# grep xhost xx__local.sh
/local/bin/set_xhost +"$(hostname --fqdn)"
/local/bin/set_xhost +localhost >
xhost +local:docker

You may want to consider install/using shellcheck to check your scripts.

On Wednesday, February 8, 2023 at 10:25:47 AM UTC+8, Bit Twister wrote:
> On Mon, 6 Feb 2023 01:40:42 -0800 (PST), hongy...@gmail.com wrote:
> > On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> >> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
> >>
> >> > Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.
> >> I suggest creating a test account, say junk. Log out of hongyi and into junk.
> >> If junk has the same problem as hongyi you know it is a system wide problem.
> >> If junk does not have the problem then you know it is a user (hongyi) problem.
> >>
> >> You can test the login process from either account by logging into the
> >> other user account by clicking up a terminal and doing
> >> su - junk
> >> from a hongyi terminal or su - hongyi from a junk terminal, depending
> >> on who you want to test.
> >>
> >> For you, I suggest anytime you make any login script change is to
> >> su - junk and very there are no problems.
> >
> > According to the above suggestions, I conducted the following test:
> >
> > werner@X10DAi:~$ sudo useradd test
> > werner@X10DAi:~$ sudo passwd test # Set the pw as: asdf1234
> > werner@X10DAi:~$ sudo usermod -d /home/test -s /bin/bash test
> > werner@X10DAi:~$ sudo mkdir /home/test
> >
> > # This time, the problem discussed here doesn't appear again:
> > werner@X10DAi:~$ su - test
> > Password:
> > test@X10DAi:~$
> >
> > So, I tried to login back as follows and the problem appeared once more:
> >
> > test@X10DAi:~$ su - werner
> > Password:
> > terminate called after throwing an instance of 'std::runtime_error'
> > what(): Failed to create dbus connection
> > xhost: unable to open display ""
> > dirname: invalid option -- 'b'
> > Try 'dirname --help' for more information.
> > readlink: invalid option -- 'b'
> > Try 'readlink --help' for more information.
> >su - werner
> > Any tips for further debugging?
> if su - test works and su - werner failed
> then some script executed by werner login is failing.
>
> No idea why
>
> Message about dirname and readlink aught to be fixed.
>
> Apparently you are running some script executing those instructions that
> test does not execute.
>
> By the way, after spending years as an maintenance programmer I found it beneficial
> to fix any errors I find even if I believe they have no bearing on the problem
> I am trying to fix.

Agreed. Nice idea!
> For your xhost problem I would need to see the xhost command. Looking through my
> host xx__local.sh script I have
> # grep xhost xx__local.sh
> /local/bin/set_xhost +"$(hostname --fqdn)"
> /local/bin/set_xhost +localhost >
> xhost +local:docker

Mine is as follows:

werner@X10DAi:~/Public/repo/github.com/bash-it/custom$ ug xhost
300-run-graphical-application-as-root-in-Wayland-session.bash: if ! xhost | grep -qE '^(LOCAL:|SI:localuser:root)'; then
300-run-graphical-application-as-root-in-Wayland-session.bash: # xhost local:
300-run-graphical-application-as-root-in-Wayland-session.bash: xhost si:localuser:root > /dev/null 2>&1
> You may want to consider install/using shellcheck to check your scripts.

Thank you for your advice.

Regards,
Zhao

On Wednesday, February 8, 2023 at 10:25:47 AM UTC+8, Bit Twister wrote:
> On Mon, 6 Feb 2023 01:40:42 -0800 (PST), hongy...@gmail.com wrote:
> > On Thursday, January 26, 2023 at 9:52:05 PM UTC+8, Bit Twister wrote:
> >> On Wed, 25 Jan 2023 23:17:06 -0800 (PST), hongy...@gmail.com wrote:
> >>
> >> > Recently, I always meet the "Are you sure you want to proceed?" message on login as described here [1], and none of the above logs give me some clues. I want to disable the above message window to appear each time when I'm trying to log in and still failed to figure out the solution.
> >> I suggest creating a test account, say junk. Log out of hongyi and into junk.
> >> If junk has the same problem as hongyi you know it is a system wide problem.
> >> If junk does not have the problem then you know it is a user (hongyi) problem.
> >>
> >> You can test the login process from either account by logging into the
> >> other user account by clicking up a terminal and doing
> >> su - junk
> >> from a hongyi terminal or su - hongyi from a junk terminal, depending
> >> on who you want to test.
> >>
> >> For you, I suggest anytime you make any login script change is to
> >> su - junk and very there are no problems.
> >
> > According to the above suggestions, I conducted the following test:
> >
> > werner@X10DAi:~$ sudo useradd test
> > werner@X10DAi:~$ sudo passwd test # Set the pw as: asdf1234
> > werner@X10DAi:~$ sudo usermod -d /home/test -s /bin/bash test
> > werner@X10DAi:~$ sudo mkdir /home/test
> >
> > # This time, the problem discussed here doesn't appear again:
> > werner@X10DAi:~$ su - test
> > Password:
> > test@X10DAi:~$
> >
> > So, I tried to login back as follows and the problem appeared once more:
> >
> > test@X10DAi:~$ su - werner
> > Password:
> > terminate called after throwing an instance of 'std::runtime_error'
> > what(): Failed to create dbus connection
> > xhost: unable to open display ""
> > dirname: invalid option -- 'b'
> > Try 'dirname --help' for more information.
> > readlink: invalid option -- 'b'
> > Try 'readlink --help' for more information.
> >su - werner
> > Any tips for further debugging?
> if su - test works and su - werner failed
> then some script executed by werner login is failing.
>
> No idea why
>
> Message about dirname and readlink aught to be fixed.
>
> Apparently you are running some script executing those instructions that
> test does not execute.
>
> By the way, after spending years as an maintenance programmer I found it beneficial
> to fix any errors I find even if I believe they have no bearing on the problem
> I am trying to fix.
>
> For your xhost problem I would need to see the xhost command. Looking through my
> host xx__local.sh script I have
> # grep xhost xx__local.sh
> /local/bin/set_xhost +"$(hostname --fqdn)"
> /local/bin/set_xhost +localhost >
> xhost +local:docker
>
> You may want to consider install/using shellcheck to check your scripts.

I wonder whether there is a similar but targeting for generating detail debug messages of startup scripts (profile, .bashrc, etc...).

Regards,
Zhao