Kubuntuer82's questions

Note 1: At the end of the post I explain why I think this is not a duplicate of this or this question.

Note 2: If you want to avoid wasting too much time in reading the question, you can just read the problem and, if you want, what I tried. The rest of the post just contains details for completeness.

The problem

In the past I upraded, by mistake, a package for which I realized later that I needed to keep the older version I used to have.

Then I downgraded the package, but apparently the newer version stayed in some cache and the aptitude tool is upgrading it periodically (every 75 minutes more precisely).

How can I prevent this?

Perhaps I should find a way to remove the new version from some cache ?

What I tried

I tried all the most common approaches to hold the package, which are described here and here, with no success.

For example, I tried using dpkg --set-selections apt-mark hold, aptitude hold and the Lock version feature in Synaptic.

In particular, the Synaptic tool even reports the package as locked, but aptitude still upgrades it, and of course I tried to hold the package using the very aptitude tool, but the same happens.

After some investigation, I realized that aptitude installs the newer version by running dpkg to install a .deb file in /var/cache/apt/archives/. Desperately, I even tried the dirty approach to just delete that file, but it is even regenerated (thus I guess it may be downloaded from the web or extracted from some tar/zip file).

Finally, I also tried to completely remove the newer version using apt-get purge and to install the older version afterwards using apt-get install and to hold it afterwards. Nothing, it is still upgraded automatically...

Details

Context

Because of compatibility issues in writing some code, after upgrading a package, let's call it thepackage, I had to change mind and I needed to downgrade it, from a newer version to an older version, say from 1.1 to 1.0.

Therefore I run something like

dpkg -i thepackage_1.0.deb

But every 75 minutes the files from this package are automatically deleted and replaced with files from the newer version 1.1, as if some system script is upgrading the package automatically.

What is happening

Then, in order to investigate about the cause, I replaced the rm tool to delete files with a script I wrote which, before calling the actual rm (which has been renamed as real_rm), prints the chain of commands which ended up calling rm (I did this by getting the PID of each process parent).

Then the log file generated by this script of mine prints something like:

COMMAND: /bin/bash /bin/rm -rf -- /var/lib/dpkg/tmp.ci

CAUSAL CHAIN:
/sbin/init
-> /usr/sbin/cron -f
-> /usr/sbin/CRON -f 
-> /bin/sh -c    /sbin/maint --mode=cron > /dev/null 2>&1 
-> /usr/bin/perl /sbin/maint --mode=cron
-> /usr/bin/perl -I /var/cache/sysmaint/production/vol/linux/ubuntu/16.04/lib/perl /var/cache/sysmaint/production/vol/linux/ubuntu/16.04/maint/017aptitude/maint --nodebug --nosilent --nodryrun --nologperline --colour --nomachineread --trace=none --mode=cron
-> /usr/bin/aptitude -o Dpkg::Options::=--force-confdef -v -y -f install
-> /usr/bin/dpkg --force-confdef --status-fd 57 --unpack --auto-deconfigure /var/cache/apt/archives/thepackage_1.1.deb

From this log, we can see that cron is periodically running aptitude by executing

/usr/bin/aptitude -o Dpkg::Options::=--force-confdef -v -y -f install

and this command executes the following subcommand

/usr/bin/dpkg --force-confdef --status-fd 57 --unpack --auto-deconfigure /var/cache/apt/archives/thepackage_1.1.deb

Therefore it seems that aptitude is being periodically executed to automatically upgrade some packages via dpkg stored in a cache folder, and one of those packages is thepackage_1.1.deb.

Note: why I think this is not a duplicate anymore.
I tried all solutions to hold the package described here and here, but they just don't work. I suspect this is a different problem because the package is already in the cache and this may be the reason why the attempts to hold the older version do not work. Perhaps I should find a way to remove it from the cache?

I know that using the v4l2loopback it is possible to create a virtual webcam or video device, which we can feed by sending streams using tools like ffmpeg.

It would be very nice to run a script automatically when the virtual webcam is switched on by an application which wants to use it.

In fact the script may run ffmpeg automatically to feed the virtual webcam each time it is switched on, instead of having to feed it manually each time.

Is this possible? What do you think?

Is there a way to redirect a real webcam to a virtual webcam from the command line, for example using the v4l2loopback module with ffmpeg ?

Details

I have an old webcam which works with most software (vlc, cheese, etc.), but not with others (Skype).

A possible workaround to make it work with such software is to redirect it to a virtual webcam using the v4l2loopback module, as done by Webcamoid, but I need to do it from the command line, for example via ffmpeg.

I am sure this is possible via ffmpeg as well, but since real webcams are rarely redirected to virtual ones, I found no examples online (the only examples I have found, like this one, redirect screen captures or existing videos to the virtual webcam instead).

Question

Do you know any tool/script to send SIGSTOP (to pause a process) or SIGCONT (to resume it) to a process by clicking on a graphical window generated by the process?

Some context

There is already a tool called xkill which kills programs by clicking on a window, but it has no options to send a particular signal. This is a pity, because somewhat related command-line tools like kill have this option.

Usually I do this via command line by manually obtaining the corresponding process names and using tools like kill, pkill or killall but it would be nice to be able to do this graphically, in a way similar to clicking on a window with the xkill tool.

Applications

I would find such a tool very useful to graphically pause some heavy processes when the CPU is too "stressed" without terminating them and without losing data which cannot be saved (e.g. for partial results of long calculations).

I need to use a browser via ssh -X (that is, X rendering is allowed via ssh), but the browser should be as lightweight as possible, since the connection is not very fast then I need to minimize the amount of data exchanged because of the GUI.

Do you know any such browser?

To make an example with editors, if I try to use an editor like kate via ssh -X with a normal (i.e. not super-fast) connection then it is extremely slow, but if I use the graphical version of emacs instead, then it becomes very good, perhaps because it doesn't need to exchange data about the GUI interface via ssh all the time (I think with emacs all the data about the GUI are downloaded just at the beginning).

What browser would you suggest for this purpose? I know there are terminal-based browsers, like links2 but I would need JavaScript support, so I cannot use it.

I know that using the sudo and ssh tools it is possible to use GUI to ask for a password in a script by setting, respectively, the SUDO_ASKPASS and SSH_ASKPASS variables in which we specify a graphical tool to ask a password.

But what if I am using any other tool which may ask for a password, for example ksu which does not seem to support any of the above-mentioned modes? (I searched on internet and I did not find anything like KSU_ASKPASS and ksu does not seem to have any option to support this).

Short version of the question

If USB ports are malfunctioning, for example they periodically disconnect the device (every 10-20 minutes), is it safe, to protect the filesystem in an external hard disk from being corrupted, to mount it read-only? Or may it still get corrupted because of electrical/power issues with the malfunctioning port?

Long version (if you are interested)

the integrated USB ports in an old laptop of mine started to be malfunctioning a couple of months ago.

In particular, every time I mount a device connected to these ports, this device randomly gets disconnected, usually after 10/20 minutes, and I have to mount it again.

Since I was using the USB ports to run Ubuntu from an external hard disk, that USB disconnection tilted the whole OS and at some point it corrupted the file system and I had to restore a previous backup of my hard disk because all the data were lost.

Then I duplicated the Ubuntu system from my external HD in the internal hard disk, in order to avoid this problem. But sometimes I still want to access the data in the external hard disk to read some data files, since I use the external hard disk at my work place.

If I mount it just as an external device and read-only does it still risk to be damaged?

I have a Linux Ubuntu 14.04 generic installed on an external SATA hard disk connected via USB 2.0 or 3.0 (depending on which PC it is connected to).

Sometimes, when I accidentally touch the cable, the USB connection between the disk and the PC apparently is affected by this, I get a lot of error messages and the whole operating system crashes.

Sometimes I even forget that I should not detach the cable and I detach it on purpose, then the same happens.

But in both cases, the problem is still the same.

Is there a way to configure the OS such that it 'freezes' waiting for the USB connection to be available again, so that when it is back the system continues the execution normally?

Especially in the last months, probably the plug on the external HD is becoming old and sometimes even just touching the cable the whole system crashes, and I risk to lose important data or even to damage the HD.

This does not happen if I mount the HD in another Linux in order to use it just to access some data files, probably because in that case the OS correctly waits for the USB connection to be stable again. Is there a way to achieve this even if the operating system is running on that very HD?

Thanks to everyone Andrea