This version provides docker wrapper v18.

https://lhcathome.cern.ch/lhcathome/result.php?resultid=432618087
First Theory doing docker.
Thank you for updating docker :-)

Proxy is set in default.local, but not used from Task.
Proxy in CentOS9 VM.
init_data.xml select info from proxy:
CVMFS_HTTP_PROXY="http://<10.116.178.201>:<3128>;DIRECT"

tested with and without pattern.

Mine seem OK

Got a proxy from the local environment
Will use it for CVMFS and Frontier
Using CVMFS on the host.
Probing CVMFS repositories ...
Probing /cvmfs/alice.cern.ch... OK
Probing /cvmfs/cvmfs-config.cern.ch... OK
Probing /cvmfs/grid.cern.ch... OK
Probing /cvmfs/sft.cern.ch... OK
Excerpt from "cvmfs_config stat":
VERSION HOST PROXY
2.13.3.0 http://s1cern-cvmfs.openhtc.io http://192.168.1.179:3128
Environment HTTP proxy: http://192.168.1.179:3128

I have without pattern per https://cvmfs.readthedocs.io/en/stable/cpt-configure.html#proxy-lists

curl -I http://example.com
answer is ok.

ChatGPT: podman squid error

ftm: cleaning it in my squid and hoping squid running correct after this.

Don't forget to set proxy in "containers.conf"

Linux podman:
etc/containers/containers.conf

[containers]

env = [
"http_proxy=http://proxy1:3128",
"https_proxy=http://proxy1:3128",
"no_proxy=localhost,127.0.0.0/8,192.168.0.0/16"
]

Impossible to set containers on hold??

Boinc manager says "Task suspended by user"
But I see in runRivet.log that crunching goes on in background.

Boinc manager says "Task suspended by user"
But I see in runRivet.log that crunching goes on in background.

I suppose that it's not in runRivet.log where you see activity, but in stderr.txt.
Only the logging (and checking) of the status of the containers goes on. The container itself is in a paused state, so no real event processing.
I admit that these logging seems much more cpu-intensive than the developers expected.
They decided to to this every 10 seconds probably for interconnection reasons with BOINC.

WOW,
CentOS9 in VM with running squid,
updated System with nearly 400 program updates.

Now Squid showing more infos in this VM.
Seem not the final step, but Docker Tasks running.....

In reply to Crystal Pellet's message of 15 Feb 2026:

I suppose that it's not in runRivet.log where you see activity, but in stderr.txt.

Checked again.
State now is "Waiting to run" because of some Atlas WUs.
But runRivet.log is updating again and again.

From your answer I conclude that it should work. So I need to find the cause.

In reply to Saturn911's message of 15 Feb 2026:

State now is "Waiting to run" because of some Atlas WUs.
But runRivet.log is updating again and again.

From your answer I conclude that it should work. So I need to find the cause.

The difference between your host and mine: Your host runs Linux OS and mine Windows OS.

Theory tasks are still failing on my machine. See https://lhcathome.cern.ch/lhcathome/result.php?resultid=432951915 for example.

BOINC tells me:
Sat 21 Feb 2026 12:23:32 PM CET | LHC@home | [error] Can't rename output file slots/4/shared/output.tgz to projects/lhcathome.cern.ch_lhcathome/Theory_2922-4763846-676_0_r998052033_result: rename() failed

It looks like the task is trying to copy/move the result file from the slots directory into the projects directory. Is it a necessary step?

---

In reply to baracutio's message of 21 Feb 2026:

It looks like the task is trying to copy/move the result file from the slots directory into the projects directory. Is it a necessary step?

Yes this is the normal step. Results are sent back to the server from the project directory.
It seems to me a permissions issue on your Linux system.

It seems to me a permissions issue on your Linux system.

ATLAS tasks (native) are running fine, other BOINC projects as well. Hard to imagine that this is a simple permission issue. Any ideas how to investigate into this problem?

---

In reply to baracutio's message of 21 Feb 2026:

It seems to me a permissions issue on your Linux system.

ATLAS tasks (native) are running fine, other BOINC projects as well. Hard to imagine that this is a simple permission issue. Any ideas how to investigate into this problem?

As you said, Atlas is running the native version, but Theory is now using docker containers.
Maybe you could use podman instead of docker.

Does this mean there won't be any virtualbox tasks anymore, and I can delete the project from my config?

In reply to Henry Loenwind's message of 22 Feb 2026:

Does this mean there won't be any virtualbox tasks anymore, and I can delete the project from my config?

ATLAS and CMS will provide VBox-tasks as it was.
Theory will send you VBox-tasks, if you have WSL disabled in Windows.

In reply to Crystal Pellet's message of 22 Feb 2026:

In reply to Henry Loenwind's message of 22 Feb 2026:

Does this mean there won't be any virtualbox tasks anymore, and I can delete the project from my config?

ATLAS and CMS will provide VBox-tasks as it was.
Theory will send you VBox-tasks, if you have WSL disabled in Windows.

Ok. I was just asking because I hadn't gotten any tasks since 19 Feb 2026, 6:58:45 UTC. It started again today at 22 Feb 2026, 7:14:53 UTC, a good while after I posted that question.

The only thing I could find in the logs was

2026-02-21 02:14:17 | LHC@home | Sending scheduler request: Requested by project.
2026-02-21 02:14:19 | LHC@home | No tasks are available for Theory Simulation
2026-02-21 02:14:19 | | Some BOINC projects require Docker.
2026-02-21 02:14:19 | LHC@home | Scheduler request completed

I started to see issues with my proxy too, i did not change anything?

Environment proxy 'http://192.168.1.179:3128' set but can't be connected
Could not find a local HTTP proxy

this is local host jsut to rule ourt any network issues.

Yes Toby,
there seem a problem with podman in combination with Squid.
Have checked with ChatGPT, but found no answer.