Aquarius Power's questions

I tried to switch to the "nvidia-driver-470" using the "Software & Updates / Additional Drivers" facility, but it always says the packages are not going to be installed.

If I select nvidia-driver-470 with synaptic, it installs all required packages, but there is something weird:
It requires an older kernel 5.4.0-1053-gcp (Google Cloud Platform), while the last one is 5.11.0-38-generic!
If I try to boot on the default option 5.11.0-38-generic, nvidia drivers wont load, and resolution will be 640x480 only!!!
If I try the 5.4.0-1053-gcp kernel (compatible with nvidia 470), it will not boot! It will stop booting when trying to initramfs, and then a few minutes later open busybox console...

Installing the 470 removed several older nvidia packages... good I had a root (and boot) backup made with rsync (https://unix.stackexchange.com/a/626503/30352), so I could quickly directly use the previously working OS copy AND/OR quickly restore it back (what I did).

I saw there are nvidia 470 for 5.11.0-38-generic packages, but I cant install them because
linux-modules-nvidia-470-5.11.0-38-generic requires
nvidia-kernel-common-470 >= 470.74, but the only available package is
nvidia-kernel-common-470 470.63.01-0ubuntu0.20.04.2
and if I download https://launchpad.net/ubuntu/+archive/primary/+files/nvidia-kernel-common-470_470.74-0ubuntu0.20.04.1_amd64.deb,
gdebi-gtk wont enable the install button for it! (I guess it is because mine is apparently 20.04.2 and the package is for 20.04.1)

but there is a report of it being installable:
NVIDIA 470 driver doesn't work after suspending Ubuntu 20.04.3 LTS (yeah, why mine with latest updates is apparently 20.04.2??)

Is 470 unstable? despite the additional drivers says it is "tested"?

I need to upgrade because UE4Editor is crashing with VK_ERROR_INITIALIZATION_FAILED "VkResult=-3" https://answers.unrealengine.com/questions/896820/out-of-memory-bug-running-422x-on-linux-with-vulka.html

I use Synaptic to upgrade my Ubuntu 18.04.

I am concerned on having problems in case of a blackout (that is not uncommon).

What I have been doing is to update at most about 20 packages at once, and always from the bigger to the smaller (I chose the Synaptic filter "Status/Installed(upgradable)"). But huge packages like more than 50MB, I update individually...

But I am still not sure how safe that is?

What is the worst thing that could happen? My machine stop booting (broken boot) in case blackout happens during an upgrade? Or is it fail proof?

And what I do is somewhat time-consuming... could it improve in some way (if there is no other better way)?

I always thought install scripts should/would continue after reboot automatically, but I never tried that (to suddenly turn off the machine) to see what happens, so?

after a whole system update (that I try to avoid usually), I thought I was unable to boot, but this time I decided to wait more.

so after exactly 6m05s (everytime is that precise delay) it finally completes booting :(

just before the update it was very fast, so I exclude any hardware problem.

[you may jump to read the last update currently 2019-12-25]

other questions I found arent related to LVM, but this case seems to be as the last failure message is about scanning for LVM:

yes, initially there is some stuff about irq and ACPI, I tried adding this but didnt help acpi=strict, so my guess is that it is a LVM problem.

I found some stuff on the log that may help (omitting "dups"):

#happens 5 times
syslog.1:Dec 18 18:52:29 MyHostName lvm[803]:   WARNING: lvmetad is being updated, retrying (setup) for 10 more seconds.

syslog.1:Dec 18 18:52:29 MyHostName systemd[1]: Created slice system-lvm2\x2dpvscan.slice.

# happens 20 times, each for a different device
syslog.1:Dec 18 18:52:29 MyHostName lvm[814]:   WARNING: lvmetad is being updated, retrying (setup) for 10 more seconds.
syslog.1:Dec 18 18:52:29 MyHostName systemd[1]: Starting LVM2 PV scan on device 8:31...

syslog.1:Dec 18 18:52:29 MyHostName systemd[1]: Started Monitoring of LVM2 mirrors, snapshots etc. using dmeventd or progress polling.

syslog.1:Dec 18 18:52:29 MyHostName kernel: [    2.460879] random: lvm: uninitialized urandom read (4 bytes read)

#logs like this happened for each device also
syslog.1:Dec 18 18:52:29 MyHostName lvm[625]:   4 logical volume(s) in volume group "MyLvmGroupName" monitored
syslog.1:Dec 18 18:52:29 MyHostName lvm[911]:   4 logical volume(s) in volume group "MyLvmGroupName" now active

The above, despite saying "for 10 more seconds" it all seems to happen on the same second at "Dec 18 18:52:29" what doesnt make much sense to me, it was many threads? anyway...

obs.: grub v 2.02-2ubuntu8.14

I check the boot time with sudo systemd-analyze time

After this latest update, when I turn on the computer, it usually "freezes" on the 1st boot (just after the boot choice is made (normal or advanced boot etc)), but I can ctrl+alt+del. After this it goes to this slowed boot problem...

19/Dec/2019 Today I saw something interesting +- like this: /dev/mapper/MyLVMGroupName: clean I couldnt take a photo, was too fast and I cant find it anywhere at /var/log (ubuntu has no boot log???) well... it appeared to me that a fsck was being run (has been being run) everytime and it is taking 6min to complete? Obs.: yesterday I did a normal shutdown, so I think it should be clean already, no need for fsck...
Another thing about this: the desktop red led (that shows us if is happening some kind of hard drive I/O) doesnt blink a single time during that 6m delay, before showing fsck clean message, so... it is not spending time reading/writing. So what is happening? could be some kind of timeout, but why it says nothing about it?

there is nothing at systemd-analyze blame that spent more than 6 seconds also.

2019-12-25: to the linux grub command, I added debug --verbose and got this!
after 60s waiting:
systemd-udevd 'SomeDevicePartition' is taking a long time
after more 120s:
systemd-udevd 'SomeDevicePartition' killed
they happen at +- : 60s, 180s, 240s, 365s
so a total of 6minutes!!!
I wonder if udevd killing timeout could be lowered to may be 10s and do not retry? (using some config in the grub entry)

My last guess is that all this is just some annoying bug :(

so, how I can fix this?

Can I hint grub to fastly detect LVM stuff?

linked: https://unix.stackexchange.com/questions/559929/very-slow-boot-6m-where-udevadm-stores-its-config/559979#559979

as I remember, everything was there on 14.04, after upgrading to 16.04 I have only these below, what can I do to fix it?

My /var/cache/apt-xapian-index has 4.5GB.

I like the synaptic search functionality a lot, but that price is too much for my small SSD...

I was thinking what I can do about it, to keep it available and also not be a disk usage issue:

• I would like to know if we can limit that size in some way?
• May be if that folder could be compressed?
• Or if it could be moved to another partition/HD, safely in a way that it would not break synaptic (or something else that also uses it) in case such partition is not mounted?
• Or some way to partially clean it, like removing information of packages that I do not care about?
• Or, may be, the xapian index could be restricted to only providing a limited indexing, so propably some data as the description of a package (as I guess it is being indexed), could be excluded from the indexing?

I am talking about the .Trash-1000 folders like.

Some times it's contents size reach like 18GB, and applications trying to write files will fail with no clear messages, because the disk space has ended.

I am looking for a way to force a trash quota/limit with some extra constraints.

These constraints options would be good:

• by file date/time, older files are erased first.
• by available disk space, if it has like less than 1GB, no matter what trash file datetime it is, it gets erased to keep 1GB available.
• by file size, if trashed file is bigger than 500MB, it is promptly erased.

Is there such software sitting somewhere so I can avoid scripting it up?

I have too many images to search visually, so I cannot open each one of them individually.

What do I have to do or install to show DDS image previews on nautilus?

I would like to preview webp too if possible.

So I have a folder I would like Nautilus to skip generating thumbnails for because there are too many files and too many changes to it.

I saw that we can disable thumbnails for a specific filetype but what about a folder?

both of these only work after the screen that was locked gets blanked; but they sometimes fail also, when for any reason the screen doesnt blanks...

gnome-screensaver-command --query
gnome-screensaver-command --time

I tried with qdbus also:

qdbus org.gnome.ScreenSaver /org/gnome/ScreenSaver org.gnome.ScreenSaver.GetActiveTime

but it equally failed.

I just found that who actually locks the screen is Unity!

qdbus com.canonical.Unity /com/canonical/Unity/Session com.canonical.Unity.Session.Lock

Related questions:
https://unix.stackexchange.com/questions/28181/run-script-on-screen-lock-unlock https://unix.stackexchange.com/questions/80143/how-to-create-a-daemon-which-would-be-listening-to-dbus-and-fire-script-on-messa

I need to track on 14.04 a crash that often happens here:

I open a new X at :1 (that sometimes goes to ctrl+alt+f8 others to f9).
But when I try to go back to X :0 with ctrl+alt+f7, it crashes about 30% of the time...

The crash happens in a way that the login screen shows up again.

I looked at /var/log/apport.log and found this:

ERROR: apport (pid 769726) Fri Aug  1 00:47:56 2014: called for pid 1457, signal 6, core limit 18446744073709551615
ERROR: apport (pid 769726) Fri Aug  1 00:47:56 2014: ignoring implausibly big core limit, treating as unlimited
ERROR: apport (pid 769726) Fri Aug  1 00:47:56 2014: executable: /usr/bin/Xorg (command line "/usr/bin/X -core :0 -seat seat0 -auth /var/run/lightdm/root/:0 -nolisten tcp vt7 -novtswitch")
ERROR: apport (pid 769726) Fri Aug  1 00:47:56 2014: is_closing_session(): no DBUS_SESSION_BUS_ADDRESS in environment
ERROR: apport (pid 769726) Fri Aug  1 00:47:56 2014: apport: report /var/crash/_usr_bin_Xorg.0.crash already exists and unseen, doing nothing to avoid disk usage DoS

I think the answer can be a generic crash tracking information about: Xorg, xscreensaver, cpufreq and other apps that could potentially cause a crash in such virtual terminal changing situation. May be also applications related to OpenGL like Unity.

I just ensured no xscreensaver opengl hacks were running, just maze; and a crash happened again.

I lock the screen with xscreensaver, but after some time the unity default screenlocker kicks in also, so I have to login two times.

Useful News:
The crash seems clearly related to Unity 3D ignoring the presence of another VT?
I saw clearly that Unity HUD gets messed like "its 3D textures" get memory corrupted?
There is this script that check and ask to replace compiz with metacity --replace.
Since I begin using it, I have not had a single crash; unfortunately, when coming back to X:0 I have to compiz --replace (the script is ready to do that also).
This other script (that opens a new X session) also does these checks/give these options.

I use ubuntu 14.04. As copy/paste for no reason suddenly fails, since nautilus on ubuntu 12.10, I am looking for a workaround.

The workaround can be any trick you came by. I have several (10+) tabs opened and there is NO WAY to close and open nautilus with all them back opened? if there is, this would be a workaround! I, at least, could find no way even with qdbus (may be I dont know how to use it very well).

In case you know what causes this problem, I would like to know too, but not required.

I was thinking, if I could collect all tab locations thru, may be, qdbus, I could create a script to re-open them on a new nautilus window.

PS.: you can help too by following @isync suggestion at comments below!

I need to know how to fix the missing "/etc/init.d/ufw" file, is it some package or some command?

I used the gufw to enable it, but on reboot my firewall was still off...

I created a symlink /etc/init.d/ufw -> /lib/init/upstart-job but I could not make it work like start ufw (neither after reboot).

I found this file /lib/ufw/ufw-init, it looks like a init.d file! can I copy or symlink it there?

Additional (optional) questions:
How to find what package has that file? apt-cache search didnt work..
Can we safely create such a script?
Any idea why is it missing?

Obs.:
My /etc/ufw/ufw.conf has ENABLED=yes (but seems useless).
iptables -L becomes different after I ufw enable, so ufw rules are not being applied on boot by any other means.

PS.: this script did NOT work either

sudo ln -s /lib/ufw/ufw-init /usr/sbin/ufw-init
sudo ln -s /lib/init/upstart-job /etc/init.d/ufw-init
#sudo update-rc.d -f ufw-init remove
sudo update-rc.d ufw-init defaults 99
sudo -k

I am trying to fastly check if an application audio is working or not (because sometimes sound fails to start. It is application bug/limitation I guess).

What I am doing is to record 1s sample and check if there is something on it. But that is slow here.

I wonder if there is some way to list all applications using pulse audio on ubuntu 12.10?

I believe it has something to do with audio listening ports; other answers pointed to that but none worked here; they only listed the pulseaudio PID, but what I need to know is: what applications are using pulseaudio?, and I need to do this in the command line to use with scripts!!

why not use tee? because the terminal rendering of the output makes the application run slower.

for some reason, this is not working:

application 2>&1 >"$logFile"

the output keeps going to terminal..

Ubuntu 12.10, gedit 3.6.1

This won't work:

gconftool-2 --type int --set /apps/gedit- /preferences/ui/recents/max_recents 30

Modifying this file as root won't work either:

cat /usr/share/glib-2.0/schemas/org.gnome.gedit.gschema.xml |grep rece -C 1

EDIT: there is nothing at gconf-editor either.

Is it hard coded? In order to increase the number of recent files in gedit from its default of 5 will I need to find, "fix" and recompile gedit with the answer here?

I am not being able to ps -p $pid of some processes that are listed at htop...

One very well known that I can point out is nautilus (so I think you can easly try it). There is a main process that ps can see, but there are 2 child processes listed on htop which pids bring nothing with ps...

how can I make ps see those child process? as some other commands like gcore can interact with them...

It is the gnome-session-properties command.

I know already the apps are at ~/.config/autostart.

I know there are files at /etc but those are root.

I found nothing at gconf-editor.

I need to know where gnome-session-properties stores what apps are disabled and enabled to auto startup, so I can backup that file :)

thx!

I am using the bash version 4.2.37 at ubuntu 12.10. But since ubuntu 12.04 I am unable to export arrays in bash...

This is an example (type these lines in the command line, it is not a script):

export astr=(a "b c" d)
declare -p |grep astr
bash
echo ${astr[@]}
declare -p |grep astr

echo outputs nothing...

declare -p |grep astr outputs nothing either...

What I am looking for is an workaround, because as far I know that is considered an known bash bug.

EDIT: btw, if possible, the workaround could avoid creating temporary storage files for the array as I may run the same script simultaneously on different shells.

I got 2 big files at, /var/log/account, of 350MB each..., my root is only 10GB...

I read there could have ckpacct to cycle and gzip it, but I cant find it, also no alternative command?

EDIT: I found that sudo accton off disable the logging but the files remain there, and I guess my next boot it will be activated again...

They seem to are being cycled but who actually does it?

369114432 May 13 23:23 /var/log/account/pacct
333708160 May 13 12:27 /var/log/account/pacct.0
 13681065 May 12 16:21 /var/log/account/pacct.1.gz
  3371433 May 11 09:50 /var/log/account/pacct.2.gz
  7549333 May 10 07:35 /var/log/account/pacct.3.gz

EDIT: my guess boot scripts does the cycle... anyway, I created this script, but I dont know what safety implications it may have.. any considerations?

cat >ckpacct.sh

#!/bin/bash

if [[ -n "$1" ]]; then
  echo "there is no parameters and no --help, read the script and understand what is does, before running it up."
  exit 1
fi

if [[ "$USER" != "root" ]]; then
    echo "you must be root to run it..."
    exit 1
fi

function FUNCerror() {
    if(($1!=0));then exit 1; fi
}

cd /var/log/account;FUNCerror $? || exit

# fast ungrab pacct file
accton off;FUNCerror $?

mv -v pacct pacct.0.temp;FUNCerror $?

echo -n |tee pacct;FUNCerror $?
chown -v root:adm pacct;FUNCerror $?
chmod -v o-r pacct;FUNCerror $?

accton on;FUNCerror $?

# compress old 0
gzip -v --best pacct.0;FUNCerror $? #releases also pacct.0 filename

mv -v pacct.0.temp pacct.0;FUNCerror $? #restore new 0 from temp

# change file names upping indexes
#mv -v pacct pacct.0
for((i=4;i>=0;i--));do
    mv -v pacct.$i.gz pacct.$((i+1)).gz;FUNCerror $?
done
rm -v pacct.5.gz;FUNCerror $? # remove last in the limit

How to see in the terminal that you have sudo privileges? (so knowing I have these privileges, I can just type sudo -k) It would be better to not add too much information into the terminal also, just some kind of simple and small alert!

REMAINING TIME IS IMPOSSIBLE, right?

Also, how to see how much time I still have with those privileges on that specific terminal? (so I can just let it end instead of typing sudo -k). I think that remaining time could show up each time I press Enter key.

EDIT: I researched more and found that /var/lib/sudo/$USER/$nPts (where $nPts is current terminal pts as of ps -p $$) has its timestamp updated everytime a sudo command is issued, and it is only accessible if you use a sudo command, so the remaining time will always be the configured one...

PS.: Alternatively to that, I tried alias sudo='sudo -k' but I dislike having to type the password for every command... And also, this alias sudos='sudo -k' (or alias ssudo='sudo -k', good if you typed a long line, just press Home and type 's') works if you only want to issue one command by typing sudos. But I still miss to know when I have normal sudo privileges to know I need to act...