LHC@home

Since 20 UTC the first task is running in SL69!
Will tomorrow see the result, because of 11 hours runtime shown.
Thank you Atlas-Team.

Since 20 UTC the first task is running in SL69!
Will tomorrow see the result, because of 11 hours runtime shown.
Thank you Atlas-Team.

You're welcome Axel

---

Volunteer Mad Scientist For Life

Thanks Magic.

Have overnight the first three Atlas SL69 finished.
BUT... only less than 50 cobblestones for the task.
Linux was at 91/8/25 the first time presentated :-).

Will now take a look to let it run in OpenSuse 42.x.

What is your Linux favorite, Magic?

Sry for late response.

Will now take a look to let it run in OpenSuse 42.x.

This will not work/you will get no native tasks because right now it HAS to be either CentOS 7 or SL6.
Once they have fixed all remaining issues it will be available to other Linux distros, but for now it is restricted to those 2. I guess they will post here once it is ready.

So far, let us see what be possible:

https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4395&postid=31906#31906

yes, i have understood this post the same way you are. but i tried it on debian (with working cvmfs and singularity) and only received vbox tasks. then i wrote here that only vbox tasks were sent, to which David replied with this:
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4396&postid=31980
But i am curious what you will report :-)

OpenSuse is only a option for the future.
Have noticed, that it is a Test-Version.
Do you have Virtualbox deinstalled for your Debian as David wrote?

We can only test step by step in the new Linux-App.

Edit: This Atlas-Tasks are part of the Production-Server.

Do you have Virtualbox deinstalled for your Debian as David wrote?

Yes. But BOINC is just complaining that Virtualbox is not installed and the server does not send anything.

Thanks Magic.


What is your Linux favorite, Magic?

Well I like to test things but since I live just NW of Microsoft Corporation I am required to say I only run Windows OS's

BUT I can't wait for the day when I can get unlimited data high-speed internet since running VB tasks use up all 80GB of my monthly total in one week so I get throttled down to about 2MBps for the last 3 weeks and Cern and VB tend to have a problem starting these VB tasks before the time limit so they end up Invalid.

It takes 3MBps or better to be sure they will all start up and get beyond HTCondor Ping

Which is why I wish we had SixTrack tasks right now since you don't have to leave all the computers online to run those tasks so once I load all mine up I can just unplug the ethernet just to make sure no data is lost (I do the same with Einstein GPU's)

So since it is 1pm here it is not easy to even start these VB tasks one at a time (same with my vLHC-dev tasks)

19 more days before I get to run at 45MBps d/l so I hope we get a new batch of SixTracks.

(and the Atlas-alpha is even harder to get running)

---

Volunteer Mad Scientist For Life

The Scratch-folder is growing to GByte.
After a few Atlas-Tasks, the message is:
There is not enough Space to Start a new Task in Boinc.
This Scratch-folder begin with ax,gb,4f......
After deleting them, it is possible to get a new task.
BUT at the moment Atlas-production-Server have ZERO tasks. :-(

By the way, Magic we hope that the ISP for you hear your words.

The scratch directory is cvmfs' cache directory. See Cache Settings. You can check your current configuration with:

cvmfs_config showconfig atlas.cern.ch
...
CVMFS_CACHE_BASE=/scratch/cvmfs    # from /etc/cvmfs/default.local
CVMFS_QUOTA_LIMIT=4096    # from /etc/cvmfs/default.local

For me, the difference between empty cache and all needed files in cache is about 40 minutes. I definitely want the files cached.

Thank you for your answer, Juha.

Have changed from 4,096 to 2,048.

After cvmfs_config reload and sudo service autofs restart,
this message is still shown,
but not every time and got also tasks from Atlas-Production Server combined with this message need more space!
OK, have Preferences changed from one task to two tasks. Maybe this is the reason.

CVMFS can be configured to use a local proxy (squid).
Thus you can keep the quota low and serve the requests from the proxy cache.

Proxy=DIRECT. Its ok so.

Have with 4 CPU's running two tasks with 2CPU's per task.
They finished after 600 sec. 0x1-Error.
Memory 10,240 GByte.

https://lhcathome.cern.ch/lhcathome/result.php?resultid=153951224

After a reboot, the same Error with ONE task and 2 CPU's.

When reload of CVMFS and default.local is changed to PROXY=DIRECT it is running.

After reboot probe and chksetup are needed to see if something is not ok.

I have one host where all my native ATLAS tasks fail with "validate error" after 6 minutes.

On my other host (my desktop PC), the ATLAS native app works fine. Both hosts are running CentOS 7 and have CvmFS and Singularity installed.

Our ATLAS friends suggested to check the home directory settings for Singularity, in fact the account running BOINC on the 2 boxes had different home directories. Now I have updated the account on the PC with the failing task with the same home directory settings as the PC that runs the native app successfully. However, it still fails.

Any further hints on this from the ATLAS team? Shall I re-install Singularity?

Hi Nils,

This computer is only running vbox WU, and I suspect the validate error is due to not enough memory in the VM. The message "FATAL makePool failed for AthMpEvtLoopMgr.SharedEvtQueueProvider" usually means running out of memory, and I've seen others say that the memory we specify for single-core tasks is not enough. I'll see if we can increase it.

This computer is running native tasks but they are failing because it looks like cvmfs is not running:

Checking for CVMFS
check cvmfs return values are 256, 256
CVMFS not found, aborting

Hi David,

Thanks, the first node used to run happily Vbox tasks, but ran out of memory as you said. It has also been running the native app successfully in the past.

I am not sure why the /cvmfs mount was lost on the second host. After re-installing CVMFS and Singularity on Tuesday, I now get a different error on this host:

Job failed: Non-zero failed job return code: 65

Ref.result: 154053235.

Boinc_conf under atlasathome have a change of the Boinc (7.5.1) at 17/8/26 6 UTC.

Have on two PC's SL69 and one run only with one CPU and not with two as defined in Preferences.

Installation of SL69 on both is identical.

This Boinc(7.5.1) is seen in the boinc_conf as old one too.

Edit: This SL69 have finished work with low cobblestones.

https://lhcathome.cern.ch/lhcathome/results.php?hostid=10496079

Job failed: Non-zero failed job return code: 65

Ref.result: 154053235.

This is the same error as the vbox tasks so I think it's a memory problem again. This machine has only 3448MB of memory, we should probably require at least 4GB for native tasks.

Have on two PC's SL69 and one run only with one CPU and not with two as defined in Preferences.

Sorry, my fault. During the Installation of the third PC with SL69 today
the Processor-use for the PC was changed from me to 95% instead of 100%.