Unexpectedly got a resend on a rig that wasn't properly set up as far as I know. It errored out after starting. Is there anything I could do to successfully run any future task I might grab?

---

In reply to Mr P Hucker's message of 23 Mar 2026:

In reply to rob's message of 23 Mar 2026:

There may be an issue when you request work from all the sub-projects. I had the same problem until I turned them all off except Theory. Whereupon I got a task that said it would run in a matter of an hour or so, but in reality is showing 100% completion (BOINC progress), is actually at about integrate 340 of 760 and just under two days, so the back of my envelope suggests another 2 days at least to completion.

This is the last work I'll be having from LHC until such time as they can clearly demonstrate that the initial estimates are about right.....

(Windows 10, vbox64:
https://lhcathome.cern.ch/lhcathome/show_host_detail.php?hostid=10787143

I'd try that, but Theory is empty.

I'd seen it recommended that "Run native if available? (Not recommended for Windows)" option in preferences be deselected when using Windows. I have it this way and haven't had problems getting tasks. Try it out, when tasks show up again. It seems to take at least a couple of days for a new batch to show up once the old one is drained.

In reply to AndreyOR's message of 24 Mar 2026:

I'd seen it recommended that "Run native if available? (Not recommended for Windows)" option in preferences be deselected when using Windows. I have it this way and haven't had problems getting tasks. Try it out, when tasks show up again. It seems to take at least a couple of days for a new batch to show up once the old one is drained.

I'll try that, but I thought it had to be selected to tell the server to give me Buddha instead of VB?

SOLVED!!!

Boinc, since it's written by the most incompetant programmer known to mankind, had set this flag when I upgraded to the latest "including docker" Boinc:

<dont_use_docker>1</dont_use_docker>

This is in cc_config.xml and needs changed to 0.

Whoops! Can anybody interpret this?


<core_client_version>8.2.8</core_client_version>
<![CDATA[
<message>
Incorrect function.
(0x1) - exit code 1 (0x1)</message>
<stderr_txt>
docker_wrapper 18 starting
docker_wrapper config:
workdir: /boinc_slot_dir
use GPU: no
create args: --cap-add=SYS_ADMIN --device /dev/fuse
verbose: 1
Using WSL distro boinc-buda-runner
Using podman
running docker command: ps --all --filter "name=^boinc__lhcathome.cern.ch_lhcathome__theory_2922-4787634-715_1$" --format "{{.Names}}|{{.Status}}"
program: podman
command output:
EOM
creating container boinc__lhcathome.cern.ch_lhcathome__theory_2922-4787634-715_1
running docker command: images
program: podman
command output:
REPOSITORY TAG IMAGE ID CREATED SIZE
EOM
building image
running docker command: build "." -t boinc__lhcathome.cern.ch_lhcathome__theory_2922-4787634-715 -f Dockerfile
program: podman
read_from_pipe() error: timeout
build_image() failed: -182

</stderr_txt>
]]>

Trying to paste some instructions but keep getting forum database errors!

Can't create post. Failed to add post to DB.

Fine..... adopt, adapt, and improve, motto of the knights of the round table. Let's try an image.

I managed to somehow get LHC to run vbox on an old laptop. Seems like Theory Simulation v301.00 (vbox64_theory) tasks run fine while Theory Simulation v302.10 (docker) ones crash after about 30 minutes of doing nothing (insignificant CPU load). Is there a way to only get vbox64 tasks? I'm not sure I can get Docker to run properly. Been struggling with the help of Gemini today, but I might just remain with ATLAS if all else fails.

---

In reply to Garrulus glandarius's message of 24 Mar 2026:

I managed to somehow get LHC to run vbox on an old laptop. Seems like Theory Simulation v301.00 (vbox64_theory) tasks run fine while Theory Simulation v302.10 (docker) ones crash after about 30 minutes of doing nothing (insignificant CPU load). Is there a way to only get vbox64 tasks? I'm not sure I can get Docker to run properly. Been struggling with the help of Gemini today, but I might just remain with ATLAS if all else fails.

Reverse how I fixed mine. In cc_config.xml, put <dont_use_docker>1</dont_use_docker>

As for getting it to work, how fast is your internet? Mine is 34Mbps, and it looks like it got impatient the first time. If you let it fail the tasks, it will eventually download enough stuff.

In reply to Mr P Hucker's message of 24 Mar 2026:

In reply to Garrulus glandarius's message of 24 Mar 2026:

I managed to somehow get LHC to run vbox on an old laptop. Seems like Theory Simulation v301.00 (vbox64_theory) tasks run fine while Theory Simulation v302.10 (docker) ones crash after about 30 minutes of doing nothing (insignificant CPU load). Is there a way to only get vbox64 tasks? I'm not sure I can get Docker to run properly. Been struggling with the help of Gemini today, but I might just remain with ATLAS if all else fails.

Reverse how I fixed mine. In cc_config.xml, put <dont_use_docker>1</dont_use_docker>

As for getting it to work, how fast is your internet? Mine is 34Mbps, and it looks like it got impatient the first time. If you let it fail the tasks, it will eventually download enough stuff.

I did the config trick recommended by Gemini, but then I kept getting red warnings that docker is not allowed and nothing downloaded. My net is slower but nowhere near as slow as some people around here (BOINC in general) complain.

Edit: weird, seems to have worked the second time around.

Re-edit: welp, it just downloaded another docker one... Running 30 minutes for no reason is pretty expensive time-wise, especially it if keeps getting many docker tasks.

Re-re-edit: Ok, this is now total black magic. It seems that the docker task is running correctly now. I can acees the running log file and it's advancing with the events.

---

Have you downloaded the latest docker auto-setup? This sets up everything correctly itself.

https://github.com/BOINC/boinc-buda-runner-wsl-installer/releases?hl=en-GB-u-fw-mon-ms-uksystem-mu-celsius

It could be yours was being impatient like mine, there is a lot of data to download. If your internet is as slow as my 34Mbit, that may have been the problem initially.

Also make sure all these are ticked (in "turn windows features on or off"):
Hyper-V
Virtual Machine Platform
Windows Hypervisor platform
Windows Subsystem for Linux

If you remove VirtualBox then you can use buda to do the docker/podman task for Theory. I set this up on a spare windows computer to test but my main windows comptuer I kept on VirtualBox for now.

Oops, looks like I forgot to mention I was talking about a Linux system, sorry guys. Since then I had 2 vbox tasks fail and the ones that are running seem to have restarted from scratch at a much slower pace... I have no idea what's going on. Guess I'll just let them run and see what works (if something finally works, lol)

Edit: I'm also very prone to being impatient so I aborted everything, restarted the rig and got a fresh batch of 8 docker tasks. The first 4 seem to be running fine now, one is relatively fast, 2 average and one very slow. All 4 CPUs got maxed out once the events started rolling so I'm trying to be hopeful. Not going to touch anything anymore.

---

In reply to Toby Broom's message of 24 Mar 2026:

If you remove VirtualBox then you can use buda to do the docker/podman task for Theory. I set this up on a spare windows computer to test but my main windows comptuer I kept on VirtualBox for now.

I have both installed. I think LHC is refusing to give me ATLAS/CMS because they would run slower, but it allows vbox from other projects (if there are any!)

It ran fine, until the end, when it crashed and has no stderr...

---

did you set up cmvfs and move podman to use cgroup's?

In reply to Toby Broom's message of 25 Mar 2026:

did you set up cmvfs and move podman to use cgroup's?

I think so. Gemini gave me some commands for setting up cmvfs and some other settings. After each crash/error, I gave the error message to the AI and it came up with apparently good solutions. Got everything working until the last step apparently (with docker). Somehow a couple of vbox tasks did finish correctly earlier, but I have no idea how to replicate those settings. I wanted to somehow filter out getting docker tasks, but it doesn't work.

---

The one you aborted seemed to be working:
https://lhcathome.cern.ch/lhcathome/result.php?resultid=434108926

the only thing in the logs was

Also add "CVMFS_USE_CDN=yes" to "/etc/cvmfs/default.local".

You might want to setup proxy on to reduce the load on the CERN servers:

https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5473
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5474


Your VBox ones vaildated so its seem like all is good.

Does cc_config.xml, put <dont_use_docker>1</dont_use_docker> not work?

In reply to Toby Broom's message of 25 Mar 2026:

... then the above comment about CDN would be not needed ...

This is wrong!

"CVMFS_USE_CDN=yes" must always be set!
Exceptions are possible for clients running inside CERN's own network or inside networks registered as CERN tier-n partners.
Most (nearly all) clients here do NOT meet that partner relationship.


In addition a local proxy (Squid) is recommended (better: it is a MUST!) for all LANs running more than 5 clients concurrently.
Typical examples:
- a multicore computer running more than 5 vbox VMs concurrently
- a multicore computer running more than 5 docker containers concurrently using CVMFS inside each container
- a mix of the previous 2 exceeding 5 concurrently running CVMFS clients
- multiple computers inside the same LAN running more than 5 VMs/containers (better: CVMFS clients) in total


What must NEVER happen:
A setup without a local proxy together with "CVMFS_USE_CDN=no".
This will cause alarms in CERN's monitoring systems and may even cause offending computers to be blocked.

Thanks for correction, sorry.