Message boards : ATLAS application : native_mt app on Ubuntu 18.08
Message board moderation

To post messages, you must log in.

AuthorMessage
[AF] Hydrosaure
Avatar

Send message
Joined: 8 May 17
Posts: 12
Credit: 11,235,723
RAC: 0
Message 36571 - Posted: 27 Aug 2018, 19:46:48 UTC

Anyone got native ATLAS app to run on Ubuntu 18.04 LTS ?

I have a Ryzen machine running this OS and am trying to get it to crunch.
cvmfs is installed and singularity 2.5.2 as well.
Still tasks fail after about 10 minutes...

Looking at the logs, it fails at this step:
check singularity with cmd:singularity exec -B /cvmfs /cvmfs/atlas.cern.ch/repo/images/singularity/x86_64-slc6.img hostname

Singularity isnt working...

running start_atlas return value is 3


Yet manually running that command it seems fine:
boinc@boinc-ab350m:/var/lib/boinc/slots$ singularity exec -B /cvmfs /cvmfs/atlas.cern.ch/repo/images/singularity/x86_64-slc6.img host
name
boinc-ab350m
boinc@boinc-ab350m:/var/lib/boinc/slots$ echo $?
0
boinc@boinc-ab350m:/var/lib/boinc/slots$


Any idea?

Host link: https://lhcathome.cern.ch/lhcathome/show_host_detail.php?hostid=10510464

Example of failed task: https://lhcathome.cern.ch/lhcathome/result.php?resultid=205978265
ID: 36571 · Report as offensive     Reply Quote
bronco

Send message
Joined: 13 Apr 18
Posts: 443
Credit: 8,438,885
RAC: 0
Message 36573 - Posted: 27 Aug 2018, 21:49:33 UTC - in response to Message 36571.  

There were problems but that didn't start until version 3. Versions 2.6.0 and 2.5.1 are working fine for me. Your version 2.5.2 should be OK and in fact it runs when typed at the command line. I was going to say it might be an execute permission problem but then noticed the command prompt suggests you're maybe logged in as the boinc user.
ID: 36573 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jun 08
Posts: 1965
Credit: 139,722,926
RAC: 86,667
Message 36576 - Posted: 28 Aug 2018, 6:06:24 UTC - in response to Message 36571.  

You may try a "cvmfs_config wipecache" and a "cvmfs_config probe".
If the last command prints errors, post them here.

Beside that you are running a 12-core setup which will try to allocate roughly 24 GB of RAM.
Not a good idea on a 16 GB computer.
You may limit the #cores per task via the project's webpage or a local app_config.xml.
A 2-core setup should be a good starting point on this computer.
ID: 36576 · Report as offensive     Reply Quote
[AF] Hydrosaure
Avatar

Send message
Joined: 8 May 17
Posts: 12
Credit: 11,235,723
RAC: 0
Message 36580 - Posted: 28 Aug 2018, 17:42:18 UTC - in response to Message 36576.  

I ran both commands; they complete correctly.
Good catch with the thread number, I've lowered it to 4 for now.

Digging some more into system logs, I've found these errors
Aug 28 18:22:12 boinc-ab350m Singularity: action-suid (U=1000,P=19539)> Home directory is not owned by calling user: /home/boinc
Aug 28 18:22:12 boinc-ab350m Singularity: action-suid (U=1000,P=19539)> Retval = 255

yet
boinc@boinc-ab350m:/var/lib/boinc/slots$ ls -ld /home/boinc/
drwxr-xr-x 8 boinc boinc 23 Aug 28 19:25 /home/boinc/

I've also tried what is listed under
http://opensciencegrid.org/docs/worker-node/install-singularity/#validating-singularity and it looks ok

boinc@boinc-ab350m:/var/lib/boinc/slots$ singularity exec --contain --ipc --pid \
>                 --home $PWD:/srv \
>                 --bind /cvmfs \
>                 /cvmfs/singularity.opensciencegrid.org/opensciencegrid/osgvo:el6 \
>                 ps -ef
WARNING: Container does not have an exec helper script, calling 'ps' directly
UID        PID  PPID  C STIME TTY          TIME CMD
boinc        1     0  0 17:32 ?        00:00:00 ps -ef
ID: 36580 · Report as offensive     Reply Quote
bronco

Send message
Joined: 13 Apr 18
Posts: 443
Credit: 8,438,885
RAC: 0
Message 36581 - Posted: 28 Aug 2018, 21:00:59 UTC - in response to Message 36580.  

Home directory is not owned by calling user: /home/boinc

That rings a bell. It looks like maybe you made the same mistake I made a while ago. Did you use gyllic's directions for building BOINC, CVMFS and singularity and later install BOINC from repos without removing the prior BOINC installation? I did and IIRC the above message turned up. After much frustration trying to sort out the permissions and ownerships I gave up and reinstalled the OS.
ID: 36581 · Report as offensive     Reply Quote

Message boards : ATLAS application : native_mt app on Ubuntu 18.08


©2022 CERN