b_laoshi's questions

I've been running Ubuntu 20.04 for some time now. I had a remote machine set up so that I could VNC into it using x11vnc at the login screen if necessary using the Xauthority file at /run/user/125/gdm/Xauthority (no, there should not be a dot in front of Xauthority in this context). All of the sudden, I try to access it today and find that I can't!

I did some digging, and it looks like the login interface was switched over from X11 to Wayland sometime in the last month or two. I could probably just switch everything back to X11, but if Wayland is the future, I'd rather find a working solution in Wayland.

Thus my questions is this. How do I get VNC working on the login screen? I do not want RDP solutions; my RDP connection is working fine!

I'm guessing the vino-server application could handle logins, but it does not start until after a user logs in. In this case, simply configuring a user to automatically login will not work. There are issues (I'm guessing with the video card driver) that cause auto logins to fail.

I'm painfully watching my apt dist-upgrade command download each package in a very long list, and what makes it painful is that packages are downloaded one-by-one. I fail to see why this process couldn't be sped up by downloading multiple packages simultaneously!

For example, if apt is downloading Chrome (PPA), Syncthing (PPA), and Ubuntu system updates (non-PPA), three simultaneous downloads will begin, one for Chrome, one for Syncthing, and one for the Ubuntu system updates.

Since all update packages are downloaded before any installation even begins, I don't see any reason why multiple packages for Ubuntu system updates coming from a single repository can't be downloaded in parallel.

I would really like to know if anyone knows a way to make this happen? Can I downloaded multiple package upgrade files in parallel from a single repository when running apt dist-upgrade or apt upgrade?

How can I determine within a shell script if the script is running:

• On a Ubuntu LiveOS (not necessarily version dependant) -- (already answered here)
• Within a chroot on an Ubuntu LiveOS (already addressed here)
• On a locally installed system (no LiveOS) -- (already answered here)

As all three points seem to be addressed elsewhere, I'm going to vote to close this question. Pointed out in the only answer here, a combination of tests can be used to determine chroot on LiveOS.

This may be the same issue as a previously asked question, but I can't say for sure because that question does not provide as many details and is for a different version and flavor of Ubuntu.

I migrated from plain old Ubuntu 16.04 with Unity to Ubuntu-Mate 16.04. For whatever reason, the network manager applet context menu no longer displays available wireless networks as evidenced by the picture below. I had no problems with this in vanilla Ubuntu.

I know it's not a driver issue because the command line interface for network manager works just fine. I am able to scan for networks, and nmcli prints a list of all of the available networks around me. Furthermore, I am able to create connections using nmcli as well. Once I have created a connection, if it is set to automatically connect, it will connect even though the applet is not listing available networks. As you can see from the image above (image was grabbed while connected to a wireless network), the gui context menu doesn't even show which network I'm connected to. It simply gives me the option to disconnect.

Does anyone have any idea how to fix this issue?

Does anyone know if it is possible to determine the status of an unattended upgrade? Is it possible to find out which packages are being installed or where we are in the process?

I appreciate the benefit of unattended upgrades, but they can be really annoying when you have a system that's been off for a couple months, and you want to install an application or when they're eating CPU resources on machines that don't have much to begin with.

I'm not simply looking to kill the unattended upgrades. I know how to do that. I want them to run. I just want to know how I can find out what's actually going on and what phase they're in.

I'm using Ubuntu Mate 16.04. I would like to configure the system so that it reboots or shuts down on logoff. Here's what I have done so far:

I created /etc/lightdm/lightdm.conf.d/50-ubuntu-mate.conf with the following content.

[Seat:*]
user-session=mate
session-cleanup-script=/sbin/reboot
allow-guest=false

While this successfully reboots the machine on logoff, I have one problem with it. It also reboots the machine when I am at the login greeter and I select a different user account for login. I do not want to reboot when I'm not actually logged into an account and all I am doing is selecting a user from the drop down to login as.

Can anyone tell me if there is another way to go about this that does not cause reboots from the login greeter?

Hypothetically speaking, say I have a system with the following configuration

• One hard drive (/dev/sda) partitioned as...
  • /dev/sda1: 25G unused space
  • /dev/sda2: 50G Windows partition
  • /dev/sda3: 25G unused space

Now I want to combine all of the unused space using LVM so that I have the full 50G available to me for my Ubuntu installation. I don't want to use /dev/sda1 and /dev/sda2 as separate 25G partitions. How can I go about configuring LVM and performing the installation so that I can run Windows and Ubuntu post-install?

How can I configure an Ubuntu Server OS (perhaps an application I can install) so that it connects to a third-party service that enables connections back into the Ubuntu Server? I'm basically looking for something like LogMeIn or VNC Connect but for Ubuntu Server rather than a Desktop interface. GUI solutions are out of the question.

I know I could accomplish this by setting up a third machine to act as an SSH gateway, but I don't really feel like managing the machine in the middle and making sure it is always online. Plus, with IP addresses that change, I don't feel like messing with DDNS or trying to keep the IP address for the machine in the middle up-to-date.

I also know I could configure the edge router to forward SSH connections to my server. I'm looking for a solution that is more independent of network architecture. The kind I wouldn't have to worry about if someone came along and pushed the reset button on the router.

For argument's sake, let's say I don't have access to the edge router. I have in fact had an ISP that placed me behind its own NAT firewall (think non-first world country); service was terrible. When I was in that situation, I couldn't set up DDNS or forward a port because I didn't have an Internet-facing IP assigned to me.

Does anyone know of a third-party service that offers the ability I'm looking for?

Let me begin by saying I'm not new to LUKS. I've set up LUKS with keyscripts numerous times with and without LVM. I'm not sure what is actually going on here though. I have a system that has a single encrypted partition. My drive is organized as follows:

# lsblk

NAME             MAJ:MIN RM  SIZE RO TYPE  MOUNTPOINT
sda                8:0    0  128G  0 disk  
└─sda1             8:1    0  128G  0 part  
  ├─vg0-root     253:1    0   20G  0 lvm   /
  ├─vg0-secure   253:6    0  100M  0 lvm   
  │ └─secure     253:7    0   98M  0 crypt /root/secure
  └─vg0-swap     253:4    0    1G  0 lvm   [SWAP]

My /etc/crypttab file looks something like this

# UUID is not required here since the path to the LV won't change
secure      /dev/vg0/secure       none      luks,keyscript=/lib/cryptsetup/scripts/insecure

My /lib/cryptsetup/scripts/insecure file is executable and looks something like this

#!/bin/sh
# My actual file looks somewhat different because it dumps the key file with dd.
# This accomplishes virtually the same thing though.

echo -n "my-encryption-password"

I have run update-initramfs -k all -u a number of times after configuring crypttab and putting my keyscript file in place.

As far as I can tell, my script file isn't even getting copied to the initrd.img file. Now that I think about it, I don't think it would get copied to the initrd.img file since the root partition is not encrypted and the script file should be easily accessible from there.

Upon rebooting, the system sees the record from crypttab and asks for a password (which in my case doesn't actually exist because the only key is a keyfile full of random bits) rather than using the keyscript to unlock the LUKS partition. I have tried taking LUKS out of the LVM and putting it on sda2, and the results were the same. I also know that the keyscript works because cryptsetup luksOpen /dev/vg0/secure secure -d - <<< "$(/lib/cryptsetup/scripts/insecure)" works like a charm and decrypts my LUKS partition.

I've tried this in Ubuntu 16.04.2 and Ubuntu Mate 16.04.2 with the same results. I've used keyscripts before without any trouble. The only difference was that, in the past, my / partition was always encrypted. If anyone can shed some light, I'd appreciate it. I only want a very small encrypted partition because I plan on cloning this system, and I don't want to clone it with the entire / partition encrypted.

UPDATE 2017-04-26

In digging through logs, I found a line with the following error which makes no sense. Since when is 'keyscript=/path/to/script' an unknown option for crypttab?

... systemd-cryptsetup[737]: Encountered unknown /etc/crypttab option 'keyscript=/lib/cryptsetup/scripts/insecure', ignoring.

Just for kicks, I tried removing the keyscript option and using a keyfile, and it all just worked! In fact, I tried other options like keyfile-offset, and they work too. Hence, the problem lies somewhere with the keyscript option. Does anyone have any idea why?