This whole adventure started when I posted in the Windows forum that one of my PC was not getting any tasks. One of the replies indicated that I might improve the VirtualBox Linux tasks if I upgraded the boinc version.

Independent from that:
Your Linux clients run BOINC v7.18.1 which is an Android only release.
Using it on Linux is known to cause lots of problems, hence must not be used.
...

Perhaps I should restore my VM to BOINC v7.18.1 which actually runs?

Well, I use 7.18.1 on all my linux boxes (the one that says pre-release) and I've not had a single problem with it. Yet I recently had problems with the Windows version 7.22 on a WINE install which worked when I dropped back to an earlier boinc version. So one might suppose there's issues with 7.22.

There's an old saying: 'If it's not broke, don't fix it' :) If I was you, I'd go back to the earlier boinc version.

I don't understand your point about not getting CPDN tasks as your profile shows you've run plenty on both the VMs? It's probably just that there are usually more Windows machines subscribed to CPDN than tasks available. I also miss out sometimes.

p.s. if you had some failed tasks lately because of the bad batch, it will take time for the server to start sending you tasks again as I believe it black-lists your host for a short while.

That rather implies that the new version of BOINC - you need to tell us where you got it from - was compiled to use a different version of the software package runtime libraries.

I knew there was a reason other than perversity for compiling my own!

Check https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/boinc

Gianfranco is an official BOINC builder.

Independent from that:
Your Linux clients run BOINC v7.18.1 which is an Android only release.
Using it on Linux is known to cause lots of problems, hence must not be used.
See:
https://boinc.berkeley.edu/forum_thread.php?id=15014&postid=112098
https://boinc.berkeley.edu/forum_thread.php?id=15014&postid=112100

After restarting the VM I got a subset of the same message:

Operation failed: read() failed

I think the problem is related to permissions as Dave says. Although your userid is now part of the boinc group, the gui_rpc_auth.cfg may not have 'group' read permission.

To fix, first note that the gui_rpc_auth.cfg in /var/lib/boinc is actually a link into the /etc/boinc-client directory:

lrwxrwxrwx 1 root root 34 Apr 20  2022 gui_rpc_auth.cfg -> /etc/boinc-client/gui_rpc_auth.cfg

and if we look at the permissions of the linked file in my files:

ls -l /etc/boinc-client/gui_rpc_auth.cfg
-rw-r----- 1 root boinc 1 Oct  1  2019 /etc/boinc-client/gui_rpc_auth.cfg

It says 'root' owns the file but the group owner is 'boinc' and we have group 'read' permission (the second 'r' in the first column). You probably don't have this set, perhaps yours is '-rw------' instead. If that's the case, then change it with:

sudo chmod g+r /etc/boinc-client/gui_rpc_auth.cfg

Hopefully that will fix it!

ls -l /etc/boinc-client/gui_rpc_auth.cfg

-rw-r----- 1 root boinc 1 Dec 2  2022 /etc/boinc-client/gui_rpc_auth.cfg

A blank gui_rpc_auth.cfg file will work - I tend to keep mine empty. There's also no special reason why /var/lib/boinc/gui_rpc_auth.cfg is a link to the default in /etc/boinc-client. You can delete the link and create a real file of the same name in /var/lib/boinc-client.

p.s. the install, I think, should really have done all this for you. Not sure why it fails on this machine and not the other you mentioned which worked.

I got something quite different:

Can't get RPC password: /var/lib/boinc-client/gui_rpc_auth.cfg exists but can't be read. See https://boinc.berkeley.edu/gui_rpc.php
Only operations not requiring authorization will be allowed.
Operation failed: read() failed

You need to add your user name to the boinc user group, and ensure that is active.

I can't remember the exact wording at this time of night, but it's been given before. It you try the boinc controls in the same terminal window afterwords, they should work: but to be safe, the easiest way is to reboot the machine - after that, the group membership will be permanent.

That should solve your problem (I forgot this step).

Check what's currently set for the boinc group with this:

grep boinc /etc/group

Mine is:

boinc:x:130:glenn

If your username is missing on the end, add it with:

sudo adduser dave boinc

and then run the 'grep' command again to check. This command will prompt you for your password.

Can't get RPC password: /var/lib/boinc-client/gui_rpc_auth.cfg exists but can't be read. See https://boinc.berkeley.edu/gui_rpc.php
Only operations not requiring authorization will be allowed.
Operation failed: read() failed

Operation failed: read() failed

Then try boincmgr/boinccmd again. You should not need to reboot the VM but I might be wrong.

If that doesn't fix it, report back. It should do.

https://boinc.berkeley.edu/gui_rpc.php 

 gui_rpc_auth.cfg 

boinc      862          1  0   11:27  ?         00:00:00  /usr/bin/boinc

Ok, so the boinc client is running using defaults. That'll mean it should be using the files in /var/lib/boinc-client and communicating on port 31416. I'm puzzled why running boincmgr in /var/lib/boinc-client didn't work in that case.

Let's see if the command line tool rather than boincmgr can talk to the client. Try this and let us know what output you get:

boinccmd --get_state

If it can see the client you'll get alot of output. If it can't, you'll get "can't connect to local host"

Can't get RPC password: /var/lib/boinc-client/gui_rpc_auth.cfg exists but can't be read. See https://boinc.berkeley.edu/gui_rpc.php
Only operations not requiring authorization will be allowed.
Operation failed: read() failed

Try that first.

For killing the client, you can use root privilege rather than as the user boinc:

sudo kill 862

but don't do that just yet.

Ok, I think most likely the client is not running but we can check. This is one of those support question that would take 5 mins to sort out if I could see the machine.

Let's see if the cilent is running and what the command line arguments are. Run this on your Ubuntu VM in a terminal (could reboot the VM first to make a clean start):

 ps -ef | fgrep 'bin/boinc'

If the boinc client is running, you'll get a line similar to this:

boinc       1404       1  0 09:56 ?        00:00:09 /usr/bin/boinc --allow_remote_gui_rpc --dir /work/boinc-client

boinc      862          1  0   11:27  ?         00:00:00  /usr/bin/boinc
dave     3570    1991  0   11:28  pts/0   00:00:00  grep -F  --color=auto  bin/boinc 

If it's not running, you'll get nothing. It's possible you might get multiple lines if, for some reason, the client has been started up multiple times (which it shouldn't). Report back what output you get. I'd like to see what command line arguments it's using. We're going to kill this later and try restarting manually, see below.

Let's also check ownership is correct. This can be another reason the client isn't starting.

Run the following command to check there is a 'boinc' user for the client:

 grep boinc /etc/passwd

should give output like this:

boinc:x:122:130:BOINC core client,,,:/var/lib/boinc-client:/usr/sbin/nologin

 boinc:x:126:133:BOINC core client,,,:/var/lib/boinc-client:/usr/sbin/nologin

Let's check the user 'boinc' owns the files in the client directory. Run this command:

 ls -l /var/lib/boinc-client

(note that /var/lib/boinc is just a symbolic link to /var/lib/boinc-client where the real files are)
You should see plenty of files, with lines like:

-rw-r--r-- 1 boinc boinc    1691 Feb  4 00:21 global_prefs.xml

The key thing is here that 'boinc boinc' part. That means the boinc user & group owns these files.

There should also be a couple of files owned by 'root' user which are links to the /etc/boinc-client directory.

If that looks ok, then first kill off any existing boinc client that's running. Referring back to the output from ps above:

boinc       1404       1  0 09:56 ?        00:00:09 /usr/bin/boinc --allow_remote_gui_rpc --dir /work/boinc-client

the number in the second column is the 'process id'. That's what you need (yours will be different). Kill the process with:

sudo -u boinc kill 1404

!! Do not use 1404 here, use the process id you got !!

Confirm the client is no longer running by using the ps command above. If for some reason you have multiple clients running, kill all of them.

boinc      862          1  0   11:27  ?         00:00:00  /usr/bin/boinc
dave     3570    1991  0   11:28  pts/0   00:00:00  grep -F  --color=auto  bin/boinc 

Now let's try starting the client at the terminal:

sudo -u boinc /usr/bin/boinc --dir /var/lib/boinc-client

(you might be prompted for your sudo password here)

This should produce output to the terminal. Then run the 'ps' command above to make sure it's working. If there are any obvious errors in the output of the command post it here.

If the client starts up ok, try the boincmgr as before:

cd /var/lib/boinc
boincmgr &

and hopefully that will connect!

I get

 19-Jul-2023 11:40:51 Another instance of BOINC is running

Ok, the client is already running. Might have started automatically when the Virtual machine was booted, or when you tried previously.

So, go into the boinc data directory and try starting the boinc manager there without any arguments:

cd /var/lib/boinc
boincmgr &

and see if you can 'Add Project -> climateprediction.net'.

 19-Jul-2023 11:40:51 Another instance of BOINC is running

If you get 'command not found', use:

/usr/bin/boincmgr &

When I run BOINC from the UI, The dialog box shows:
BOINC Manager (Pre-release) - Communication
Communicating with BOINC client. Please wait

When I try to start the client from the terminal, I get "command not found".

As you are on Ubuntu in virtualbox you may not have the systemctl running, but you can start the boinc client manually just to test it.

On Ubuntu, the client executable is:

$ ls -l /usr/bin/boinc
-rwxr-xr-x 1 root root 989552 Jan 30  2022 /usr/bin/boinc

Is your 'boinc' executable in /usr/bin? If not, the install put it somewhere else. Try looking in the installer log to see what directory it put it into.

If boinc is there, what do you get if you type:

 /usr/bin/boinc --version

(if your boinc is somewhere else, change the directory name appropriately.

I get:

7.18.1 x86_64-pc-linux-gnu

 7.20.5 x86_64-pc-linux-gnu

If you get similar or same, then try starting the client with:

/usr/bin/boinc --dir /var/lib/boinc

and report back what output you get.

 19-Jul-2023 11:40:51 Another instance of BOINC is running

Also, have a look in the /var/lib/boinc directory and see if you have a file called 'stdoutdae.txt'. This is the logfile for the client.

If that exists, then the client started up. Run the command:

tail -20 /var/lib/boinc/stdoutdae.txt

and send us the output and we can help.

Some things BOINC only looks at during client startup. If OpenCL drivers are added a client restart is needed for the client to see the drivers. Was the client restarted after the VM disk size change?