Déjà vu's questions

It happened twice in ~2 months, since I replaced Gnome with Unity (ubuntu 22.04.1).

While watching a YT video in Chrome, and Unity just froze. The video was looping (audio only), repeating the last 2 seconds forever. Mouse and keyboard dead. Crl-Alt-Fx not working. ssh to the desktop unresponsive. Had to reset the computer.

Both times, /var/crash showed that the culprit seems to be glade2script-python3.

The man says that

Glade2script allows a script (Bash, Python or other) to display and interact with a GUI (created with Glade).

Can I safely remove glade2script-python3? Is it useful?

I run VS-code and did "Add to favorites" to start it easily.

However, I also need to have the -disable-gpuoption, otherwise VS-code hangs forever.

But doing "Add to favorites" Gnome did only save the Exec path, not the options that were accompanying it.

Is there a way to change the Dock launcher to add this parameter?

In other words, is there a Dock Launchers config available somewhere that allows that kind of change? (didn't see it in dconf)

_{NB I know it's possible to create a new .desktop file, then "run/start" it, then Add to favorites. My questions aims directly at the Dock config. Thanks}

tl;dr Looking for a reliable alternative to Thunderbird that

• can import (or read) all emails from Thunderbird (only emails is fine)
• support local account (read localhost mail, from /var/mail/...)
• option: has clean configuration files (not like prefs.js)

Longer shot:

after upgrading from Ubuntu 18.04 to 20.04, the newer Thunderbird 91+ doesn't support reading local emails anymore.

Not only that, but the Inbox that was linked to that "local" email account is not showing anymore (data is still here).

I tried messing again with the (oh my...) pretty-mezzy prefs.js, but, this time... giving up.

TB's recommendation is to create a local POP (or IMAP) server, but, if possible, I'd rather do a migration to something reliable, clean, and certainly lighter, that reads local mail directly.

Time for a change.

Thanks.

Got this from the TB folks

https://forums.mozillazine.org/viewtopic.php?t=3086614

And for the time being, just did

Just to clarify to other people, you only need to install mailutils-pop3d and restart your system. In Thunderbird just add your mail as a normal mail account, user@localhost and your login password"

(and actually that was straightforward ; apt install and TB account config)

Still opened to recommendations, though.

History (the question is in 2020 below)

2018. After the fresh install of 18.04, two years ago, I wanted to use my /etc/mysql/my.cnf setup from 16.04 to get things back to what they were. But the MySQL install sets two symlinks

• /etc/mysql/my.cnf -> /etc/alternatives/my.cnf
• /etc/alternatives/my.cnf -> /etc/mysql/mysql.cnf

thus I replaced the new /etc/mysql/mysql.cnf with my old 16.04 /etc/mysql/my.cnf.
Things went fine until another MySQL update, which, for some reasons, had my mysql.cnf replaced with the new one. Bug? Did I accept the replacement? Don't think so, but, well, that's always possible.

So what I did at the time was to bypass the alternatives system, i.e. replace the /etc/mysql/my.cnf link with my my.cnf file, and zeroed mysql.cnf. Since my.cnf takes precedence over mysql.cnf that worked (I think I tried to symlink mysql.cnf to my my.cnf file, and there were other trouble, don't remember).

(To be honest, didn't like much the alternatives implementation that I found, cumbersome and counter-intuitive)

2020. Did an upgrade from 18.04 to 20.04. And naturally, MySQL re-created a clean environment with the alternatives links. Good.

However, to complete the upgrade, this time I'd like to keep things clean, and as much as possible

• keep using the alternatives system or not (depending on your advice)
• ensure that a further MySQL upgrade (unless it's v. 9) won't overwrite my setup (or it asks)

What is the recommended way to keep a clean MySQL install?

The question is about the console you get by hitting CtrlAltFx (F2 for instance). Not the Terminal from a gnome (...) session.

Whatever the size of the font I choose, the bottom line is always lacking ~2 pixels, thus the cursor is not visible.

I tried to modify /etc/default/console-setup and add

SCREEN_HEIGHT="800"

instead of "900" to see if anything changes... but nothing changes.

Is there a way to either

• Change the console number of lines (I'd set the current number - 1)
• Change the screen height to allow for the cursor to be visible

(preferred solution would be the 'number of lines' one)

It's Ubuntu 18.04 on a Macbook pro 15" mid-2015, installed as an alternative boot (using refind). fbset output

mode "2880x1800"
    geometry 2880 1800 2880 1800 32
    timings 0 0 0 0 0 0 0
    accel true
    rgba 8/16,8/8,8/0,0/0
endmode

I'm rather glad of the upgrade from 16.04 to 18.04 (desktop).
However there are a few things I find annoying,

Among them is indicator-multiload

indicator-multiload is an invaluable app, graphically showing activity for CPU, memory, network, disk... Looking like that (top bar) in 16.04

Unfortunately it doesn't work (well) in 18.04.

How can I get indicator-multiload to work in 18.04?

Tried to add fcitx-mozc module to input Japanese on 17.10

apt install fcitx-mozc

then set the input method to fcitx ▾

But for some reason "日本語 Mozc" does not appear in the Input Sources (Text Entry).
(this after trying several times, rebooting in between)

It seems being on 17.10 makes a difference.

Is there another module to install? Or maybe 17.10 does not support fcitx-mozc?

Now that systemd took over the services management, the power controlling programs are a link to systemctl

chezmoi# ls -l /sbin | grep systemctl

lrwxrwxrwx  1 root root 14 Jul 19 08:56 halt -> /bin/systemctl*
lrwxrwxrwx  1 root root 14 Jul 19 08:56 poweroff -> /bin/systemctl*
lrwxrwxrwx  1 root root 14 Jul 19 08:56 reboot -> /bin/systemctl*
lrwxrwxrwx  1 root root 14 Jul 19 08:56 runlevel -> /bin/systemctl*
lrwxrwxrwx  1 root root 14 Jul 19 08:56 shutdown -> /bin/systemctl*
lrwxrwxrwx  1 root root 14 Jul 19 08:56 telinit -> /bin/systemctl*

The command reboot in particular will shutdown all services to restart them later on, but without going to a warm machine restart, like it used to do during the init era.

• is there a way to reboot down to the BIOS from the command line?

Could be also a setting I've set a long time ago that I forgot ...

^{Using Ubuntu 16.04}

Edit:

• Doing a reboot, the desktop session ends, the Ubuntu splash screen appears and runs for a few seconds, then 3 lines appear furtively: "kdump loaded", "kdump unloaded", "Starting new kernel", then it's back to the login page.

• That happens while doing the reboot command, from a terminal, as root

Downloaded the Ubuntu 16.04.2 .iso install disk and flashed it onto a USB stick.

Booting on the USB stick, there is no memtest86+ option (along with "Try Ubuntu").

Is there a way to add it to the install options?

(it's for a UEFI boot PC running normally Windows, its owner does not want to install Ubuntu. "memtest86+" is completely free (unlike "memtest86"), but is quite old, and does not offer a UEFI boot - this is why having it within the Ubuntu options makes things easier)

Trying to recover a bunch of .desktop files from Ubuntu 14.04 copied to 16.04, but they don't appear in the Launcher.

Did the following (/14.04 is mounted to the old Ubuntu disk / FS)

cd ~/.local/share/applications
cp -pr /14.04/home/ring0/.local/share/applications/[bunch-of].desktop .

but they do no appear after logging back to 16.04. (same user / uid on both versions)

Had a look at /usr/share/applications, but the files are in the .local dir.

Looking at the structure of the .desktop files on both versions, they look similar.

Anything else to check / setup?

Thanks

After installing Ubuntu 16.04 on a brand new computer, I tried to copy all Google Chrome data (Extensions, Bookmarks, Passwords, ...) from my old 14.04.

Especially extensions, since there's a lot of them!

Both 14.04 and 16.04 have the same user name (ring0) and ID (1000).

Copied (cp -pr as root [while the files belong to ring0 anyway])

~ring0/.config/google-chrome
~ring0/.cache/google-chrome
~ring0/.gnome/apps/chrome*

and even

 /opt/google

after a proper install (just in case a key is stored somewhere)

But, while the bookmarks have been recovered automatically (thanks to the login to Google Gmail et al.),

• Have to log back in to all sites (eg this one)
• No extension has been copied

Since I'm root on both machines, and have access to "Show passwords" in the settings anyway, is there a way to copy and get everything, including extensions, without having to login again?

On my system (14.04), I'm not using www-data as the owner of nginx and php-fpm, using web instead.

^{(for the curious: 1. because www-data is a ugly name, 2. for not using a user that everyone knows is the default).}

AFAIK, all necessary settings have been changed accordingly (while www-data is still a valid user in /etc/passwd)

• nginx setup
• php-fpm setup
• /etc/logrotate.d/nginx doing create 0640 web adm instead

Doing a grep -r www-data /etc/* shows only some lines for the user files (passwd, shadow...) - nothing else.

Yet, every now and then, for some reason, access to /var/log/nginx is back to www-data:adm.

It seems to be related to logrotate though its settings have been changed, and the grep shows no relevant match.

Is there something else to look to ensure /var/log/nginx will be set to web:adm for good, and not set back to www-data:adm some time in the future?

In Unity launcher (icons on left side), is it possible to configure a .desktop file to, for instance, run something else if Control is pressed.

For instance, Gimp below

The file gimp.desktop in /usr/share/applications shows (excerpt)

[Desktop Entry]
Version=1.0
Type=Application
Name=GIMP Image Editor
GenericName=Image Editor
Comment=Create images and edit photographs
Exec=gimp-2.8 %U
TryExec=gimp-2.8
Icon=gimp
Terminal=false

A click on the icon runs gimp-2.8 %U.

Is there a way to add some configuration to that file in order for a Control-click (for instance) to run another command, gimp-2.8 --no-cpu-accel %U for instance?

It's on a laptop running Ubuntu 14.04, where the WiFi wlan0 is used to connect to Internet.

And eth0 (RJ45) is set statically in /etc/network/interfaces

auto eth0
iface eth0 inet static
    address 192.168.11.168
    netmask 255.255.255.0
    gateway 192.168.11.100

The problem is, there is usually no cable connected to eth0, so at boot the IP address is not set. And when a cable is connected (on a LAN) for a short period of time, the address has to be set

ifdown eth0
ifup eth0

and since the device the laptop connects to does stop and restart, the IP is lost and has to be set again.

That kind of configuration does not seem to fit for a laptop connecting at times to devices, which may reboot etc...

Is there a way to tell the system to keep that IP assigned to eth0, whatever happens, cable disconnections, boot, etc... ?

edit

It seems wicd was the culprit. Once I removed eth0 from its config, the IP address behaved more consistently. It's always ON, on boot, unplugged... I knew the good old interfaces file could be trusted!

Ok so it took some time, but was able to install and use teamviewer on a remote 14.04, at last.

This post was of some help.

Basically (that part is actually an answer by itself!)

• install the 32 bits package version (14.04 is multiarch)
• start the daemon, set a password (see link above)
• get the ID from /opt/teamviewer/config/global.com
• start the client as regular user

and after a few core dumps (harmless seemingly) of the daemon (Linux binary) while setting password, ranting about TV and their lack of option consistency (and no error when entering a wrong option), ... having to start the client (using wine since no bin client exist seemingly) for some reasons (while being remote).. (All of that question the quality of TV developments, but that's not the point...)

And that's it - or... is it?

Well, had to figure out that the error "Authentication rejected" kept coming on to the (actual) client (a Mac) - the logs being useless - because the client (server side!) was expecting a License Agreement OK click... (used xdotool, gnome-screenshot, rsync to visualize a screenshot of the Ubuntu screen!)

Finally using xdotool could manage to remote-click the OK (license agreement) + the next warning... and hurra! Worked!

For the sake of completeness, efficiency and serenity... when a local install was not possible/made prior to remote access

• is there a way to install teamviewer without all these pirouettes - i.e. at least without having to locally agree the License Agreement (which defeats the purpose of a remote install - xdotool is a nice tool but that's really awkward!) ?

Have a VirtualBox Ubuntu guest (14.04) used for the sake of testing only.
The root user has its own password, so I can login as root in a separate terminal (su -), or directly from a login session (eg after Ctrl-Alt-F4).

$ su -
Password: <current root password>
# echo Works!

Since it is a VB testing system, and the root user does not need any security, I want to be able to su - or login as root quickly, i.e. without root having a password.

So, tried as root

# passwd -d root

to remove the root password - it seems to work according to /etc/shadow (::)

root::16304:0:99999:7:::

but when trying to su -, it asks for a password, I just press enter (no password)

$ su -
Password: 
su: Authentication failure

but it doesn't su.

There has to be a setting somewhere to allow root to login/su without password.
Where would that be?

Ubuntu 12.04, Gnome classic.

Trying to unbind/change < Altclick-1 > that moves windows (reason being another app is using that and anyway I want to change it).

But couldn't find it in

• How to unbind Shift-Alt Up shortkey in 12.04? (compiz)
• How do I unbind Ctrl-Alt-T? (config mgr)

Compiz has "Move window" disabled, and I don't find anything relevant in config-manager to remove that key binding.

Any idea?

Ubuntu 11.10 desktop up to date (Gnome 3)

To prevent answers in haste, please notice that

• GRUB_CMDLINE_LINUX_DEFAULT="" is set in /etc/default/grub
• Tried also GRUB_CMDLINE_LINUX_DEFAULT="acpi=off"

This removes most of the Ubuntu splash screen and shows most of the messages.

However, just after the BIOS launches the Operating system, Ubuntu starts by "preparing" the splash: the screen is ubuntu-colored for about 1 second (plain color, looks somewhat like the default color after a login). During this 1 second some messages are displayed that cannot be seen.

I wouldn't care much, but unfortunately during 50% of the boots, and during this very second, Ubuntu asks something... don't know what since I cannot see the messages. So I do Control-D (tried ^D once and it worked, so I keep doing it again) and the messages page is back and immediately cleared (the messages before the ^D can be seen for a split second - i.e. can't see them) and Ubuntu resumes the boot normally.

Unfortunately dmesg doesn't show what was asked during that second. Nothing in the logs. But I'd like to know what happens. So the question:

Is there a way to prevent that pre-splash colored page that hides the messages for 1 second to appear?

Just upgrade to 11.10 (from 11.04). Waited a couple of months to get a "safer" version.

Unity comes by default - ok.

Since I cannot stand Unity, being an old-time Ubuntu user and having a few applets / shortcuts on the top bar (like cpu,mem monitoring, start a bunch of apps...), I did

 apt-get install gnome-session-fallback

and chose Ubuntu Classic in the login menu. Works.

But... the top bar is now almost empty, all my icons disappeared, no more monitoring... Not mentioning the admin menus ...

How can I recover all the icons and menus from my before-upgrade era?

edit

And also, where can I find the system monitoring applets for Gnome 3? (the 11.04 gnome 2 ones do not work anymore)

I just installed PlayOnLinux, it says:

  An updated version of PlayOnLinux is available (3.8.11).

and nothing more.

Doing apt-get update doesn't offer any updated version of POL.

What is the safe / regular way to update playonlinux?