knb's questions

I have "only" a 64 Gigabyte root partition, /, hence implicitly the /var dir is on this partition, and so is snap's multi-GB subdir /var/lib/snapd/snaps . Snap stores the snap images in this directory. This takes quite some disk space.

I have freshly installed these snaps: boxy-svg, core, core18, core20, core22, eclipse, fx, gnome-3-34-1804, gnome-3-38-2004, go, gtk-common-themes, inkscape, minizinc, postman, powershell, rustup, slack, snapd, tidy, youtube-dl, zoom-client, and they already need 2,7G for a single-instance install. For backups this will increase to 10 GB quickly, because snap stores images of the 2 previous images too, in the long run.

I want to move these 10 GB on /var/lib/snapd/snaps to a different partition, on a different SSD. How can I do that properly?

I've tried by

1. just by stopping snapd with sysctl,
2. move /var/lib/snapd/snaps to a different partition
3. symlinking to that partition ln -s /mnt/big-disk/snaps /var/lib/snapd/snaps
   1, restart snapd

However, this construct does not "survive" reboots (snaps cannot be mounted during startup), and also the snaps cannot be updated easily. As a workaround I've tried scripting commands such as
mount /var/lib/snapd/snaps/inkscape_10426.snap /snap/inkscape/10426
(with /var/lib/snapd/snaps/ being a symlink), but these are a nightmare to maintain, because the release ids such as 10426 always change.

I have some apps that do not displayed correctly in the dock of my ubuntu 20.04 which I run on a standalone desktop PC.

The icon is shown with 100% transparency. That means the icon is there, it occupies some space, when the app has is active (?), there is a little orange dot-indicator on the left margin of the icon.

All is well, except that the icon is invisible

These screenshots illustrate what I see and what I expect to see.

Actual	Expected

Any idea how to get the icon displayed correctly?

Maybe it is important to mention that this logo is from the "Slack" app, which was loaded as a Snap:

which slack
/snap/bin/slack

Name Version Rev Tracking Publisher Notes
slack 4.26.1 62 latest/stable slack✓ -

and that the Slack-App does have a little Status-Indicator in the top right corner of the top menubar.

When I try to run performance monitoring and reporting tool "dstat", I simply type

dstat

on the command line. I don't have it installed. Ubuntu 20.04 comes up with this output:

> dstat
# command output

Command 'dstat' not found, but can be installed with:

sudo apt install dstat  # version 0.7.4-6, or
sudo apt install pcp    # version 5.0.3-1

Which package is "official"? Is pcp actually better? It has the higher version number, whereas dstat is v0.7. - Or does pcp simply include dstat 0.7. but also includes a web frontend (or something like that)

Get some basic info about both tools:

dstat

apt info dstat | grep -A 2 Description:

Description: versatile resource statistics tool
 Dstat is a versatile replacement for vmstat, iostat and ifstat.

pcp

apt info pcp | grep -A 2 Description:

Description: System level performance monitoring and performance management
 Performance Co-Pilot (PCP) is a framework and services to
 support system-level performance monitoring and performance

Update: Meanwhile I have installed dstat. I still do not understand why Ubuntu proposes both packages, as equivalent alternatives (or).

I want to extend Inkscape (installed with snap) by putting the SVGO-Optimizer SVGO in Inkscape's local extension directory.

According to the Inkscape preferences dialog (Edit/Preferences/System), the directory to use is (and to put extensions in)

$HOME/.config/inkscape/extensions/

There I want to install inkscape-svgo.inx. I followed the inkscape-svgo instructions.

however, Apparmor does not let me execute the extension from within the Inkscape GUI. I cannot execute the "Save As SVGO-Optimized SVG " dialog.

An uninformative "Saving as [any filename] failed" dialog appears.

I suspect it is an Apparmor issue.
How can I defined or extend the Apparmor rule to allow Inkscape to execute the SVGO extension?

journalctl gives me this output:

AVC apparmor="DENIED" operation="open" profile="snap.inkscape.inkscape" name="/proc/148064/mountinfo" pid=148064 comm="inkscape" requested_mask="r" denied_mask="r" fsuid=1000 ouid=1000
AVC apparmor="DENIED" operation="exec" profile="snap.inkscape.inkscape" name="$HOME/.config/inkscape/extensions/inkscape-svgo" pid=149456 comm="inkscape" requested_mask="x" denied_mask="x" fsuid=1000 ouid=1000
Script::execute(): failed to execute program '$HOME/.config/inkscape/extensions/inkscape-svgo'. 
Reason: Failed to execute child process “$HOME/.config/inkscape/extensions/inkscape-svgo” (Permission denied)

My Inkscape version:

snap list inkscape
Name      Version                    Rev   Tracking       Publisher  Notes
inkscape  1.1-ce6663b3b7-2021-05-25  9090  latest/stable  inkscape✓  -

I have installed Ubuntu 20.04 LTS on a new computer, and decided to use the Logical Volume manager lvm .

The installer partitioned the hard drive into 3 partitions, (1 MB UEFI bootlader area, , then I assigned 1 GB as to /boot and 150 GB to / which is /dev/sda3.

The output of sudo parted /dev/sda print was

Model: VMware Virtual disk (scsi)
Disk /dev/sda: 161GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags: 

Number  Start   End     Size    File system  Name  Flags
 1      1049kB  2097kB  1049kB                     bios_grub
 2      2097kB  1076MB  1074MB  ext4
 3      1076MB  161GB   160GB                      

Now I after executing

sudo parted /dev/sda set 3 lvm on

(because I can) , I get

Number  Start   End     Size    File system  Name  Flags
 1      1049kB  2097kB  1049kB                     bios_grub
 2      2097kB  1076MB  1074MB  ext4
 3      1076MB  161GB   160GB                      lvm

Which is better? Is The lvm flag really important to have (or to not have)? Did the Ubuntu installer *not set this flag on purpose?

The Logical Volume Manager (lvm) metadata daemon seems to be available (and recommended to be in a running state) in Ubuntu up to version 18.04 LTS, but for 20.04 it is not even available

sudo systemctl status lvm<tab> does not list it.

on 20.04:

$ sudo systemctl status lvm2

lvm2-activation-early.service  lvm2-monitor.service
lvm2-activation.service        lvm2-pvscan@8:3.service
lvm2-lvmpolld.service          lvm2.service
lvm2-lvmpolld.socket 

but on 18.04:

lvm2-activation.service   lvm2-lvmpolld.service     lvm2-pvscan@8:16.service
====================
lvm2-lvmetad.service      lvm2-lvmpolld.socket      lvm2-pvscan@8:5.service
lvm2-lvmetad.socket       lvm2-monitor.service      lvm2.service
===================

(highlighted by me)

Why? is this documented somewhere?

I can find a forum post on the ArchLinux Wiki, but I'd like to find something more "official". Has lvm2.metad been replaced with something? Or merged?

In Ubuntu 18.84, there is some documentation available on the LVM metadata daemon, but for 20.04 and higher it is not. An equivalent manually-crafted link to 20.04 currently redirectly to 18.04.

For editing screengrabs, I prefer to use ksnip, since Shutter was no longer installable without hassles on Ubuntu 18.04 and 20.04

However ksnip is available as a Snap App on my 20.04

○ → snap list | grep ksnip
ksnip                    1.8.0                       265    latest/stable    dporobic               

Unfortunately I have to start ksnip from the command line, then use the File-Open Dialog to pick a screengrab/PNG, then edit it in ksnip. This is should be easier.

What I want is to use Nautilus File manager, then select the "Open with other Application" context menu item to select ksnip, and the have ksnip open the file I picked with nautilus.

Nautilus Context Menu, selecting a PNG:

"Select Application" Dialog:

How can I achieve this?

There is a "View all Applications" Button, but after clicking, ksnip does NOT appear in this list of apps. "Find new applications" does not work either.

Maybe this is a duplicate, but I think we did not have the combination yet (Hack the dialog and Snap App)

(Note to self: maybe Shutter is usable again. Development had stalled for a while, or it needed a lot of deprecated versions of gnome libs as dependencies) .

I have an Ubuntu 16.04 Server that runs

Linux 4.4.0-187-generic #217-Ubuntu SMP Tue Jul 21 04:18:15 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux, all packages are very recent (I regularly perform standard updates). There is not much installed on that server: apt list --installed | wc -l yields 889 Packages,

and the filesystem has these partitions:

Filesystem                           Size  Used Avail Use% Mounted on
udev                                 2,0G     0  2,0G   0% /dev
tmpfs                                396M   26M  370M   7% /run
/dev/mapper/host--vg-root        45G  6,8G   36G  16% /
tmpfs                                2,0G   88K  2,0G   1% /dev/shm
tmpfs                                5,0M     0  5,0M   0% /run/lock
tmpfs                                2,0G     0  2,0G   0% /sys/fs/cgroup
tmpfs                                2,0G     0  2,0G   0% /run/shm
/dev/sda1                            472M  133M  315M  30% /boot
/dev/mapper/host--vgdata-data1  1,8T  824G  958G  47% /data1
tmpfs                                396M     0  396M   0% /run/user/1000

The system tries to convince me to upgrade to 18.04.5, but I am hesitant. I'd like to upgrade for consistency reasons, because that is the version that I have installed on a few other machines.

If the upgrade process fails I would be in trouble.

Anyone tried that specific upgrade ? (I'm asking because here someone reports problems upgrading from 14.04 to 20.04. There are more posts like this.)

[Update: Please follow the instuctions in the accepted answer. A PPA, snap or flatpack is not necessary if you install from the source code.]

Is there a PPA that has ugrep executables available for Ubuntu 18.04?

The Github repo only has links for Debian .deb files for various architectures, and I cannot install that on Ubuntu 18.04. The ugrep release available on packages.debian.org is v2.1 - quite old (current is 2.5) but for Ubuntu 18.04 this still seems to be too new:

sudo dpkg -i ugrep_2.1.1+dfsg-1_amd64.deb 

(Reading database ... 912390 files and directories currently installed.)
Preparing to unpack ugrep_2.1.1+dfsg-1_amd64.deb ...
Unpacking ugrep (2.1.1+dfsg-1) over (2.1.1+dfsg-1) ...
dpkg: dependency problems prevent configuration of ugrep:
 ugrep depends on libgcc-s1 (>= 3.0); however:
  Package libgcc-s1 is not installed.
 ugrep depends on libstdc++6 (>= 9); however:
  Version of libstdc++6:amd64 on system is 8.4.0-1ubuntu1~18.04.

dpkg: error processing package ugrep (--install):
 dependency problems - leaving unconfigured
Processing triggers for man-db (2.8.3-2ubuntu0.1) ...
Errors were encountered while processing:
 ugrep

ugrep tagline:

NEW ultra fast grep with interactive query UI: search file systems, source code, text, binary files, archives (cpio/tar/pax/zip), compressed files (gz/Z/bz2/lzma/xz/lz4), documents, fuzzy search, and more. A faster, user-friendly replacement for GNU/BSD grep. )

I'm on Ubuntu 18.04.

When I do this:

ll /usr/share/backgrounds/

I see this link:

lrwxrwxrwx ... ubuntu-default-greyscale-wallpaper.png -> Beaver_Wallpaper_Grey_4096x2304.png

Why is this link to a Default Grey Wallpaper there?

I have some ideas: Perhaps it is used during Hibernation or Suspend, or when a Ubuntu runs as a Virtual machine and is "Paused" .

Anyone?

Open-source security tool "tiger" sent me an email containg this:

The configuration file /boot/grub/menu.lst has world permissions. Should be 0600

This is reported twice, once as a warning and in the same mail on the next line, as a failure.

Now I have two questions:

• Should I set permissions as the tool recommends? (This is for a fileserver exposing only very few ports, but to the general internet)
• Is this a misconfiguration of the tool (because it occurs twice)?

When I run the command sudo systemctl list-unit-files (I think the sudo is optional), I get output which shows all services and their state.

Here is a snippet from my machine:

UNIT FILE                                  STATE
...
debian-fixup.service                       static  
debug-shell.service                        disabled
display-manager.service                    enabled 
dns-clean.service                          enabled 
dsmcad.service                             enabled 
emergency.service                          static  
failsafe-x.service                         static  
friendly-recovery.service                  masked  
fuse.service                               masked  
gdm.service                                masked  
getty-static.service                       static  
[email protected]                             enabled 
gpsd.service                               indirect
[email protected]                           static  
gpu-manager.service                        enabled 
halt-local.service                         static  
halt.service                               masked  
hostname.service                           masked
...

I wonder why some services are in the "masked" state. I think this means, "this is better than 'disabling', because the service cannot be started, neither by hand or by systemd".

How can I get more information about the state of a service unit?

Who has put the units into their respective state?

I tried, for example, sudo systemctl help dsmcad - that only brings up the documentation = ... line from the unit file. /etc/systemd/system/dsmcad.service

Note: Here I know exactly what dsmcad service is and what it does, I have installed it myself. I am more interested in a general solution.

I have thermald running on my desktop PC, which is an older Xeon with 4 cores.

thermald pollutes the /var/log/syslog file on my 15.04. (See this related question, How to quiet Ubuntu 15.04 noisy syslog)

It creates entries like this every 4 seconds:

Jun 23 16:01:08 xxx thermald[30886]: Set : 89000, 40000, 0, 0, 0

I could figure out how to change the config file to make it less noisy, so I turned thermald off completely (temporarily).

sudo systemctl stop thermald

Is this safe? Should I also blacklist the kernel module?

I have some problems with the desktop-search tool "recoll", which comes with ubuntu. (I have 15.04)

Actually, there are several subquestions, but maybe they are all related. I suspect all questions can be solved with an upgrade of the software and a rebuild of the document-index.

I have a recoll index that has grown to 18 Gigabyte file

du -sh  ~/.recoll/xapiandb
18G     /home/knb/.recoll/xapiandb

This directory contains several subdirs and files.

When the recollindex process runs, it runs for days at 100% CPU, it never stops until I kill it with kill -9 PID. Why is that so? How to diagnose which files/dirs it is working on, or thrashing on? Research on the internet gives inconclusive answers. Would rebuilding the index solve the 100%CPU problem?

Is it good practice to exclude this directory from my enterprise backup software (over which I have not much control)?

Should I delete the index from time to time in order to make it more effective? Or, after dropping, would it simpley build the same 18-GB index, or, worse, an incomplete index?

Should I upgrade recoll to the newest version?

Recoll version: Recoll 1.17.3 + Xapian 1.2.19 - default for 15.04.

Wily werewolf , 15.10 will contain Recoll 1.20 anyway.

(according to https://launchpad.net/ubuntu/+source/recoll)

I've just found these directories on my ubuntu PC. How do they come about?

/var/tmp/systemd-private-762c8e6ced154b18ac76720c2c05f2aa-colord.service-fcAgPb
/var/tmp/systemd-private-762c8e6ced154b18ac76720c2c05f2aa-rtkit-daemon.service-iVKzv4
/var/tmp/systemd-private-809e24ac9abd4087b9941b3590575e64-colord.service-KlzdX8
/var/tmp/systemd-private-809e24ac9abd4087b9941b3590575e64-rtkit-daemon.service-mNk1of

The directories each only contain another subdirectory tmp which is always empty.

I mean the filename is kind of self explanatory. systemd probably creates them and they have to do with an rtkit and colord services/demons.

But why does systemd create them? Are these services misconfigured?

Some of these directories are older. Some have a modification time of the last boot.

Can I delete them?

I wonder if it is possible have Nautilus Gnome/Unity configured to always pre-populate the location area in the "open file" or "save as" dialogs .

I want to pre-populate it with the current directory. Copying the current directory to the clipboard See attached screen shots.

Current state: (on 14.04):

What I want to achieve:

Nautilus Main Window does prepopulate the location field with the path:

I know there exists a possible workaround with the nautilus-open-terminal extension, but that is not what I'm asking here.

I've just had a look inside the directories /etc/init.d and /etc/rc.d . I have a 13.10 machine, upgraded from 13.04.

Then I saw that this link was missing.

S99acpi-support -> ../init.d/acpi-support

What is this script suppposed to do?

Maybe it was migrated to another location?

I have other files with similar names on this machine. None of them seems to be a shell script. Should I get it from elsewhere, e.g. from the default repos? Alternatively, should I just remove the invalid links?

locate /acpi-support
/etc/default/acpi-support
/usr/share/acpi-support
/usr/share/acpi-support/state-funcs
/usr/share/doc/acpi-support
/usr/share/doc/acpi-support/README
/usr/share/doc/acpi-support/changelog.gz
/usr/share/doc/acpi-support/copyright
/var/lib/acpi-support
/var/lib/dpkg/info/acpi-support.conffiles
/var/lib/dpkg/info/acpi-support.list
/var/lib/dpkg/info/acpi-support.md5sums
/var/lib/dpkg/info/acpi-support.postinst
/var/lib/dpkg/info/acpi-support.preinst

UPDATE Aug 2017: The solution marked as accepted by me apppears to have worked until Ubuntu 15.10. For 16.04+ choose one of the more recent solutions.

For 16.04+ the config file is /etc/bluetooth/main.conf NOT /etc/bluetooth/audio.conf .

I have a bluetooth speaker, Sony SRS-BTX300, which works in linux/ubuntu 13.04, but only after some fiddling. (Update Dec 2013: same problems remain in 13.10)

I have to set the preferred bandwidth mode to "High fidelity playback" (A2DP) each time after switching on the computer and reconnecting the speaker.

The mode resets itself to "telephony/duplex" (=low bandwidth) every time. It takes about 20 clicks to reset the speaker, (click through menu, disconnect, reconnects, select mode, test), and these are definitely too many clicks.

How can I make "A2DP" mode the default mode, an do so in a persistent manner?

A few screen shots will illustrate.

After choosing this menu by clicking on the bluetooth icon in the top panel in Unity...

After setting up this...

... the preferred mode will be reset to this after restarting/suspending the computer

Ubuntu 12.04 is out, and I want to upgrade. However, I need to conserve my existing perl 5.12 installation before upgrading. On 12.04, there is perl 5.14 and I would need to reinstall lots of perl modules. I better copy my existing perl installation before ugrading, because the installer will remove the old perl stuff and I cannot determine what I need when the 5.12 modules are gone.

Any ideas how to do this? I have a mix of modules installed from repos, from cpanm, from cpan, and from source. I need

• a strategy to determine a comprehensive list of my perl modules, their respective origins
• a sequence in which order to proceed.
• advice on what to do in case a certain module is not yet available in 5.14.
• advice on which environment variables and paths to change.

Maybe this belongs to Serverfault or launchpad.net, but here we go anyway:

In Oneiric, there is an export command in this script /etc/apache2/envvars , setting LANG. This gets ignored on my server. Any idea why?

I had two more commands in Script /etc/apache2/envvars

export SYBASE=/opt/sybase  
export SYBASE_OCS=OCS-15_0

but they get ignored as well as the output from the printenv.cgi script shows.

Is this a bug or a security feature in Oneiric?

Note that the answer from Sep 27 2011 at 17:29 to this question explicitly says that admins should add export commands to /etc/apache2/envvars .

I found this: A quote from the Oneiric changelog regarding Package sbox-dtc

- Removed the use of Apache envvars which aren't safe (easy to hack).