I upgraded linux mint to v21 about a month ago. I've been T/S off and on since but have reached a point of giving up.

lm-bhyve ~ $ /etc/init.d/boinc-client status
× boinc-client.service - Berkeley Open Infrastructure Network Computing Client
Loaded: loaded (/lib/systemd/system/boinc-client.service; enabled; vendor preset: enabled)
Active: failed (Result: core-dump) since Tue 2022-09-27 11:09:58 EDT; 2s ago
Docs: man:boinc(1)
Process: 2545 ExecStart=/usr/bin/boinc (code=dumped, signal=ILL)
Process: 2552 ExecStopPost=/bin/rm -f lockfile (code=exited, status=0/SUCCESS)
Main PID: 2545 (code=dumped, signal=ILL)
CPU: 59ms

Sep 27 11:09:57 lm-bhyve systemd[1]: Started Berkeley Open Infrastructure Network Computing Client.
Sep 27 11:09:58 lm-bhyve systemd[1]: boinc-client.service: Main process exited, code=dumped, …=4/ILL
Sep 27 11:09:58 lm-bhyve systemd[1]: boinc-client.service: Failed with result 'core-dump'.

Linux lm-bhyve 5.15.0-48-generic #54-Ubuntu SMP Fri Aug 26 13:26:29 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

boinc-client/jammy,now 7.18.1+dfsg-4 amd64 [installed]
boinc-manager/jammy,now 7.18.1+dfsg-4 amd64 [installed]
boinc/jammy,jammy,now 7.18.1+dfsg-4 all [installed]
libboinc7/jammy,now 7.18.1+dfsg-4 amd64 [installed,automatic]

Checking resources for these installed items shows no missing dependencies, etc. I've tried to reinstall, also.

stderrgui.txt shows some information but I can't make much sense of it. I don't see where I can attach that log file but here is the gist.

(boincmgr:14531): Gdk-CRITICAL **: 07:05:44.129: gdk_window_thaw_toplevel_updates: assertion 'window->update_and_descendants_freeze_count > 0' failed
(boincmgr:23569): Gtk-CRITICAL **: 19:06:33.874: gtk_box_gadget_distribute: assertion 'size >= 0' failed in GtkScrollbar

I realize this is for manager and not the client but I can't find any other logs.

Where did you get those various systemd and boinc-... files from? Which repository?

Ign:1 http://packages.linuxmint.com vanessa InRelease
Hit:2 http://packages.linuxmint.com vanessa Release
Hit:3 http://security.ubuntu.com/ubuntu jammy-security InRelease
Hit:4 http://ppa.launchpad.net/kisak/kisak-mesa/ubuntu jammy InRelease
Hit:5 http://archive.ubuntu.com/ubuntu jammy InRelease
Hit:7 http://archive.ubuntu.com/ubuntu jammy-updates InRelease
Hit:8 http://archive.ubuntu.com/ubuntu jammy-backports InRelease
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
All packages are up to date.

Using "apt"

systemd came with the boinc-client install
https://boinc.berkeley.edu/wiki/Installing_BOINC_on_Ubuntu
--Creates the client "init" script, /etc/init.d/boinc-client, and a set of configuration options for the client at /etc/default/boinc-client

There's something suspicious about the versions of BOINC v7.18.1 included in those repos - we haven't got got to the bottom of it yet.

Could you try removing those versions, and try these instead?

sudo add-apt-repository ppa:costamagnagianfranco/boinc
sudo apt update

Gianfranco Costamagna's PPA is a fast-track link for recent BOINC releases, automatically compiled from Debian sources. The current version, v7.20.2, is running fine on my Mint 20.3 - I'd be interested to hear how it runs on Mint 21 - that would help us distinguish between a problem in Mint 21 / Ubuntu 22 themselves, and a problem in the BOINC packages specific to those OSs.

Edit - source reference for the suggested PPA:
https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/boinc

Hot damn! That did it...I'm back in business.

Thank you!

Great. I see that Mint have released an upgrade tool for converting v20.3 installations like mine to v21. I was planning to reduce my computing load from 01 October anyway (when my electricity supplier increases their prices yet again), so I'll take a crunching timeout and try it out.

My working hypothesis for the problem you encountered rests on one simple fact: BOINC v7.18.1 was never released for the desktop operating systems Apple, Linux, Windows. It was only ever released for Android. I suspect the Ubuntu repo maintainers saw the Android release notice, and pulled a broken source tree. Now I just need some help in proving it, and examples like this help me down that route.

Very good! I hope my experience will help in that endeavor. Thank you again!

Note: I had reviewed the release history...but it didn't even dawn on me it was an android version!
https://boinc.berkeley.edu/wiki/Release_Notes

Great. I see that Mint have released an upgrade tool for converting v20.3 installations like mine to v21. I was planning to reduce my computing load from 01 October anyway (when my electricity supplier increases their prices yet again), so I'll take a crunching timeout and try it out.

My working hypothesis for the problem you encountered rests on one simple fact: BOINC v7.18.1 was never released for the desktop operating systems Apple, Linux, Windows. It was only ever released for Android. I suspect the Ubuntu repo maintainers saw the Android release notice, and pulled a broken source tree. Now I just need some help in proving it, and examples like this help me down that route.

If all that is true, then ask the folks at openSUSE what they did to make it work -- because 7.18.1 does work here, quite well.
I am having one problem -- I cannot access the VM Console in the Boinc manager, and I have tried just about everything I can think of to make it work.

boinc-client runs as a systemd service when target multi.user (aka runlevel 5) is reached (it is not enabled by default, so must first be enabled: su - systemd enable boinc-client). When boinc is installed, it creates a user and user group, both named "boinc". The client runs as user boinc, and to enable the manager to communicate with it, it is necessary to add yourself to group boinc. Additionally, if VirtualBox is installed, the boinc user must be added to group vboxusers (created when VBox is installed) -- otherwise, boinc-client cannot create a virtual machine.
All this works just fine, except as noted above.

7.18.1 was working fine for me on one of my systems (a Docker image, running on Linux x86_64). but the maintainer/automation has since upgraded it to 7.20.2 now.

---

Well, this is very odd. I finished off all work, set 'no new tasks' on every project, reset every project, stopped and disabled the systemd service, and took a deep breath.

Then I updated Mint v20.3 (as advised), and launched the upgrade tool for Mint 21.

Many hours, and a good lunch, later, it went through the final grumbles and declared itself complete. Reboot.

More grumbles on the pre-launch text screen, but it loaded fully. I still had a BOINC icon on the desktop, so I clicked it. And it was all there, manager and client, all running - including detection of my NVidia GPUs and their drivers.

Some oddities:

Sat 01 Oct 2022 16:08:50 BST |  | Version change (7.16.6 -> 7.18.1)

I was running 7.20.2 before, and I haven't run 7.16.6 for years.

Then it went into a mode I've complained about before.

Sat 01 Oct 2022 16:14:21 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:15:24 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 864.00 seconds; 1.00 devices
Sat 01 Oct 2022 16:16:26 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 432.00 seconds; 0.50 devices
Sat 01 Oct 2022 16:17:32 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:18:36 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:19:38 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:20:40 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:21:41 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:22:43 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:23:46 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:24:48 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:25:50 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:26:52 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:27:54 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:28:56 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:29:59 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:31:00 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices

I had an absolutely minimal work request set (0.01 days + 0.00 additional). So why does it ask for (and get - Einstein is a very consistent project) the full amount every 60 seconds - even though it's got it already? It's as if it doesn't register all the previous successful requests. Grumble.

It went on asking and getting while I was typing, until

Sat 01 Oct 2022 16:43:22 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:43:24 BST | Einstein@Home | No work is available for Gamma-ray pulsar binary search #1 on GPUs
Sat 01 Oct 2022 16:43:24 BST | Einstein@Home | see scheduler log messages on https://einsteinathome.org/host/12808716/log

I'll go and see what that log says.

No, nothing except the 'out of work' message. And it's gone back into the endless fetch'n'fill:

Sat 01 Oct 2022 16:53:20 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:54:30 BST | Einstein@Home | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices

And now it's doing the same for GPUGrid:

Sat 01 Oct 2022 16:57:37 BST | GPUGRID | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:57:38 BST | GPUGRID | [sched_op] estimated total NVIDIA GPU task duration: 1440557 seconds
Sat 01 Oct 2022 16:58:13 BST | GPUGRID | [sched_op] NVIDIA GPU work request: 1728.00 seconds; 2.00 devices
Sat 01 Oct 2022 16:58:14 BST | GPUGRID | [sched_op] estimated total NVIDIA GPU task duration: 1440557 seconds

GPUGrid is only allowed to run on one of my GPUs - that remained in place:

Sat 01 Oct 2022 16:08:50 BST | GPUGRID | Config: excluded GPU.  Type: NVIDIA.  App: all.  Device: 0

I would probably upgrade to Gianfranco's offering and see what happens.

Yup, that's probably tomorrows job. I've just allowed the second GPUGrid task to run on the second GPU, because they like their work back quickly. That'll probably keep it busy overnight, and then I'll take stock. Luckily, I upgraded the RAM on that machine, ready for the CPDN IFS trials, so it should cope comfortably.

But as a diagnosis for why the client/manager won't launch on Mint 21 (see thread title): I'm afraid it was an abject failure.

Well, that was fun - not.

Fished off yesterday's overfetch, without any drama. Then removed v7.18.1, with a view to installing Gianfranco's PPA. That didn't work - there appears to be a new key security policy in place for Mint 21 (I've reported it to him).

Then I put v7.18.1 back in place - and it wouldn't work, either. I had to add myself to the BOINC group all over again, before the GUI password could be read. Another reboot, and at last I was back in business. Seems to be behaving itself now.

But I still haven't got a clue why other people are having so much trouble. The keys seem to be:

• Learn the systemctl commands for systemd (stop, start, enable, disable)
  
• Add your user account to the boinc group
  
• reboot
  

But I still haven't got a clue why other people are having so much trouble. The keys seem to be:

• Learn the systemctl commands for systemd (stop, start, enable, disable)
  
• Add your user account to the boinc group
  
• reboot
  

Glad to see things have returned to what they should be for you. Hopefully it will stay that way from now on.

I would make a couple of small changes to your list items:

Learn the systemctl commands
Add your user account to group boinc, and if using VirtualBox add user boinc to group vbox-users
Log out/log in to complete your user's addition to group boinc (in openSUSE at least, this does not appear to be necessary if you use YaST's user manager to do this task -- try it with your distro's equivalent to see if it works for you)
Instead of rebooting, simply enable boinc-client service: sudo systemctl enable boinc-client
And now start boinc-client: sudo systemctl start boinc-client

I fixed mine by removing all Boinc from Software Manager and using do all the Mint updates.
go back into Software manager and go to bottom of list something like edu.berkely.boinc sort of looked different than all other Boinc apps, I installed it.
if you don't see it in Software manager go to package manger and see it is in there.
The only thing it is not doing now is running the client with nvidia GPU stuff. In other words it uses CPU only.
But it is a much older nVIdia card with an old legacy driver that I always have trouble with Mint, but it is a play/experiment machine.
However GPU is not showing up in other machines with newer cards that I also upgraded to Mint 21.
It seems in Mint 21 that Boinc runs in sandbox or flatpak not sure,, haven't had the time to investigate any more, to many irons in the fire, but Boinc runs, just not GPU progects.

---

Win11 x64-Asus AMD 2.7GHzX4-16GB-GT730 2GB
Garuda x64-ASUS-Intel 2.7GHzX2-16GB-GT450
Mint21.3 x64-Xeon 2.4GHzX4-32GB-nVidia GT730 2GB
Garuda Linux x64-Intel-2.1GHz-8GB
OMV-7 x64-NASIntel-i3-3.6GHz-32GB-29TB

Both my machines - currently Mint 21 / BOINC 7.18.1 and Mint 20.3 / BOINC 7.20.2 - run NVidia GPUs just fine. You just have to take time and trouble over choosing and installing the right drivers.

Hi folks,

Boinc crashes too in my machine with Ubuntu 22.04.1, both recently installed

I got boinc with sudo aptitude install boinc-client boinc-manager

Maybe there is another thread addressing my problem, but I didn't see it

Please be polite, I am novice at linux and this forum, and English is not my mother tongue. Thanks!

Logs:

syslog

Oct 22 17:21:04 x systemd[1078]: Started Application launched by gnome-shell.
Oct 22 17:21:04 x kernel: [14174.584702] traps: boincmgr[10628] trap invalid opcode ip:556c3cb1455d sp:7ffe4f187c70 error:0 in boincmgr[556c3cae8000+19c000]


apport.log

ERROR: apport (pid 10631) Sat Oct 22 17:21:04 2022: called for pid 10628, signal 4, core limit 0, dump mode 1
ERROR: apport (pid 10631) Sat Oct 22 17:21:04 2022: executable: /usr/bin/boincmgr (command line "boincmgr")
ERROR: apport (pid 10631) Sat Oct 22 17:21:05 2022: debug: session gdbus call: (true,)
ERROR: apport (pid 10631) Sat Oct 22 17:21:06 2022: wrote report /var/crash/_usr_bin_boincmgr.1000.crash


_usr_bin_boincmgr.1000.crash

ProblemType: Crash
Architecture: amd64
CurrentDesktop: ubuntu:GNOME
Date: Sat Oct 22 17:21:04 2022
DistroRelease: Ubuntu 22.04
ExecutablePath: /usr/bin/boincmgr
ExecutableTimestamp: 1643531387
ProcCmdline: boincmgr
ProcCwd: /var/lib/boinc-client
ProcEnviron:
PATH=(custom, no user)
XDG_RUNTIME_DIR=<set>
LANG=es_ES.UTF-8
SHELL=/bin/bash
...
ProcStatus:
Name: boincmgr
Umask: 0002
State: S (sleeping)
Tgid: 10922
Ngid: 0
Pid: 10922
PPid: 1239
TracerPid: 0
Uid: 1000 1000 1000 1000
Gid: 1000 1000 1000 1000
FDSize: 512
Groups: 4 24 27 30 46 122 134 135 1000
NStgid: 10922
NSpid: 10922
NSpgid: 1239
NSsid: 1239
VmPeak: 214384 kB
VmSize: 214384 kB
VmLck: 0 kB
VmPin: 0 kB
VmHWM: 44140 kB
VmRSS: 44140 kB
RssAnon: 7860 kB
RssFile: 36280 kB
RssShmem: 0 kB
VmData: 3100 kB
VmStk: 132 kB
VmExe: 1648 kB
VmLib: 102132 kB
VmPTE: 260 kB
VmSwap: 0 kB
HugetlbPages: 0 kB
CoreDumping: 1
THP_enabled: 1
Threads: 1
SigQ: 0/63303
SigPnd: 0000000000000000
ShdPnd: 0000000000000000
SigBlk: 0000000000000000
SigIgn: 0000000001000000
SigCgt: 0000000000000000
CapInh: 0000000000000000
CapPrm: 0000000000000000
CapEff: 0000000000000000
CapBnd: 000001ffffffffff
CapAmb: 0000000000000000
NoNewPrivs: 0
Seccomp: 0
Seccomp_filters: 0
Speculation_Store_Bypass: thread vulnerable
SpeculationIndirectBranch: conditional enabled
Cpus_allowed: 3
Cpus_allowed_list: 0-1
...
Mems_allowed_list: 0
voluntary_ctxt_switches: 2
nonvoluntary_ctxt_switches: 45
Signal: 4
Uname: Linux 5.15.0-52-generic x86_64
UserGroups: adm cdrom dip lpadmin lxd plugdev sambashare sudo
CoreDump: base64
H4sICAAAAAAC/0NvcmVEdW1wAA==
...

I finally purged this version and installed that of ppa:costamagnagianfranco/boinc, and no crashes anymore, but I had two problems, fixed after quite googling and testing:

1) boinc client did not start at booting: I finally changed the name of the initialization script /etc/init.d/boinc-client, then

update-rc.d boinc-client remove

then

update-rc.d "new-name" defaults

where "new-name" is another name for the initialization script. Reboot and boinc client finally starts

2) boinc manager did not connect to boinc client, fixed with

xhost si:localuser:boinc

Then I created a simple script to execute that command and then launch boinc manager, as xhost returns to the previous state at rebooting (i.e. to simply avoid writing the command each time).

Another thing to do is what is told here:

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

I hope this work could be useful to someone, and of course I'm opened to any improvement.

Richard: thanks