Development for LHC@home

This new version provides the latest docker wrapper from upstream.

I tested with three different options:

1. https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3654852 "Run test applications in preferences selected": Got a docker task running
2. https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3655038 "Run test applications in preferences not selected": Got a docker task running
3. https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3655443 "WSL Windows disabled with bcdedit /set hypervisorlaunchtype off": Got a vbox task running

@Laurence: your Linux tasks give only errors: https://lhcathomedev.cern.ch/lhcathome-dev/results.php?hostid=5328 with couldn't start app: Can't write init file:
Do you have the latest BOINC version?

Thanks for testing. My failures are on my laptop which I was using for testing. I should disable it. We plan to upgrade the production server on Monday and will release docker for Theory shortly afterwards.

I found an anonymous volunteer running Windows 11 with only errors with this new application.
He/She reached the quota, so only getting 1 task a day now.
The tasks seem to start OK, but ending about 10 minutes after start with
read_from_pipe() error: timeout
build_image() failed: -182

Example: https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3656793

The whole result log:
<core_client_version>8.2.8</core_client_version>
<![CDATA[
<message>
Incorrect function.
(0x1) - exit code 1 (0x1)</message>
<stderr_txt>
docker_wrapper 17 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__lhcathomedev.cern.ch_lhcathome-dev__theory_2922-4865519-469_0$" --format "{{.Names}}|{{.Status}}"
program: podman
command output:
EOM
creating container boinc__lhcathomedev.cern.ch_lhcathome-dev__theory_2922-4865519-469_0
running docker command: images
program: podman
command output:
REPOSITORY TAG IMAGE ID CREATED SIZE
docker.io/library/almalinux 9 c9bcec02f046 2 weeks ago 196 MB
EOM
building image
running docker command: build "." -t boinc__lhcathomedev.cern.ch_lhcathome-dev__theory_2922-4865519-469 -f Dockerfile
program: podman
read_from_pipe() error: timeout
build_image() failed: -182

</stderr_txt>

Edit:I found another client running Windows 11 and his docker tasks version 7.65 are running OK: https://lhcathomedev.cern.ch/lhcathome-dev/results.php?hostid=5442&offset=0&show_names=0&state=0&appid=4

Further testing this new docker version with Windows 10 and BOINC version 8.2.8:

1. Suspend task with and without keeping application in memory: After resuming the task, the task continues where it left off.

2. Stop and start BOINC client without suspending the tasks before: The tasks continue where they were left off.

3. Reboot the machine: All docker tasks starting from the very beginning.

In reply to Crystal Pellet's message of 24 Jan 2026:

3. Reboot the machine: All docker tasks starting from the very beginning.

:-(

On the Applications page the application

Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU	7.65 (docker)	22 Jan 2026, 13:14:40 UTC	0 GigaFLOPS

keeps staying on 0 GigaFLOPS

Also the Linux version, but I didn't check whether someone did run that.

In reply to Crystal Pellet's message of 25 Jan 2026:

Also the Linux version, but I didn't check whether someone did run that.

https://lhcathomedev.cern.ch/lhcathome-dev/results.php?userid=616&offset=0&show_names=0&state=0&appid=4 is running 7.65 on Linux. Various results, can't explain why some errors out and others are succesfull.

In reply to boboviz's message of 24 Jan 2026:

In reply to Crystal Pellet's message of 24 Jan 2026:

3. Reboot the machine: All docker tasks starting from the very beginning.

:-(

I'm running Tft on my notebook
I restated it after 23hrs and the wus restated from 0%. After 8hrs i need to restart the pc again and, again, restarted from 0%.
I killed these 2 wus...

It is great that scenarios 1 and 2 are working. Scenario 3 should be reported upstream in the BOINC github issue tracker.

In reply to Laurence CERN's message of 27 Jan 2026:

Scenario 3 should be reported upstream in the BOINC github issue tracker.

I chatted on X, some days ago, with Vitalii Koshura, a boinc developer, about this and he said that they are working on checkpoint for docker (up to now without great success). But they don't give up!

In reply to boboviz's message of 24 Jan 2026:

In reply to Crystal Pellet's message of 24 Jan 2026:

3. Reboot the machine: All docker tasks starting from the very beginning.

:-(

A little solace when you don't need a real reboot: Stopping BOINC and stop the PC in hibernation mode: The tasks survive this and continue where they were left off.

In reply to Crystal Pellet's message of 27 Jan 2026:

A little solace when you don't need a real reboot: Stopping BOINC and stop the PC in hibernation mode: The tasks survive this and continue where they were left off.

I know how to suspend the calculation.
But sometimes it is impossible for me....

When a Theory-docker task is suspended (not Leave Application In Memory=LAIM) or BOINC-client is restarted, the state of the task is saved properly in the container,
but when the task is resumed, BOINC-client starts counting the CPU- and elapsed time from 0 (zero)

Example: https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3672368

from BOINC task:
Run time 1 hours 33 min 40 sec
CPU time 45 min 37 sec

from JOB:
job: run exitcode=0
job: diskusage=106176
job: logsize=76 k
job: times=
0m0.009s 0m0.026s
851m38.330s 22m29.565s
job: cpuusage=52448
Job Finished

This also affects BOINC's credit calculation.

Bug reported at Github.

In reply to Crystal Pellet's message of 29 Jan 2026:

When a Theory-docker task is suspended (not Leave Application In Memory=LAIM) or BOINC-client is restarted, the state of the task is saved properly in the container,
but when the task is resumed, BOINC-client starts counting the CPU- and elapsed time from 0 (zero)

Retested: Suspending alone does not affect the cpu- and elapsed times. Only BOINC stop and start does.
Probably due to the absence of checkpoints.