Message boards : Theory Application : Installation of CVMFS
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38292 - Posted: 19 Mar 2019, 12:41:56 UTC

Well, unfortunately I'm a bit of a Linux noob.
I tried to install CVMFS according to the guide on a Virtual Box-Linux with openSuse 13.1 but I couldn't get it to work.
I tried the Scientific Linux commands to install it but the yum-command seems to be unknown to my Suse.
Also the Debian/Ubuntu ones don't work also, dpkg is also unknown it seems.
Is it possible at all to run this on openSuse?
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38292 · Report as offensive     Reply Quote
computezrmle
Avatar

Send message
Joined: 15 Jun 08
Posts: 1133
Credit: 55,676,700
RAC: 100,593
Message 38294 - Posted: 19 Mar 2019, 13:30:20 UTC - in response to Message 38292.  

... Is it possible at all to run this on openSuse?

Yes. I'm still running 13.1.

You may try the CernVM-FS Core Packages for openSuSE 13.2 from the download page:
https://cernvm.cern.ch/portal/filesystem/downloads

You also need the latest CernVM-FS Config Package from the same page.
Install both packages with rpm, zypper or yast.


Then follow the instructions at:
https://cernvm.cern.ch/portal/filesystem/quickstart
https://cvmfs.readthedocs.io/en/stable/cpt-quickstart.html


Important hint:
It makes no sense to request native BOINC task from LHC@home before CVMFS works, i.e. "cvmfs_config probe" returns "OK".
The tasks require apps and data from the CVMFS.
ID: 38294 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38336 - Posted: 20 Mar 2019, 7:53:37 UTC

Well, thanks for the tips but I couldn't get this damn CVMFS installed at all no matter what I tried meanwhile. :-\
I'm too inexperienced with Linux to dig into this even more and don't have the time either so I give up here.
Too bad... I hope a Windows app for this will come out sooner or later, if not then so be it...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38336 · Report as offensive     Reply Quote
mmonnin

Send message
Joined: 22 Mar 17
Posts: 44
Credit: 3,801,950
RAC: 10
Message 38368 - Posted: 21 Mar 2019, 22:05:52 UTC - in response to Message 38336.  
Last modified: 21 Mar 2019, 22:06:03 UTC

Well, thanks for the tips but I couldn't get this damn CVMFS installed at all no matter what I tried meanwhile. :-\
I'm too inexperienced with Linux to dig into this even more and don't have the time either so I give up here.
Too bad... I hope a Windows app for this will come out sooner or later, if not then so be it...


If you want to try Ubuntu you can use the instructions I gave to my team. It's worked for me on several 18.04 installs. The instructions given at LHC forums are way overly complicated for no reason when CVFMS can be installed from a repository. It's also much quicker than compiling it.

https://www.overclock.net/forum/365-overclock-net-boinc-team/1721752-projects-month-march-2019-a.html#post27876964
ID: 38368 · Report as offensive     Reply Quote
Profile Laurence
Project administrator
Project developer

Send message
Joined: 20 Jun 14
Posts: 290
Credit: 227,111
RAC: 19
Message 38373 - Posted: 22 Mar 2019, 9:55:47 UTC - in response to Message 38368.  

Well, thanks for the tips but I couldn't get this damn CVMFS installed at all no matter what I tried meanwhile. :-\
I'm too inexperienced with Linux to dig into this even more and don't have the time either so I give up here.
Too bad... I hope a Windows app for this will come out sooner or later, if not then so be it...


If you want to try Ubuntu you can use the instructions I gave to my team. It's worked for me on several 18.04 installs. The instructions given at LHC forums are way overly complicated for no reason when CVFMS can be installed from a repository. It's also much quicker than compiling it.

https://www.overclock.net/forum/365-overclock-net-boinc-team/1721752-projects-month-march-2019-a.html#post27876964


Our instructions for the Theory app point to the official CVMFS documentation. Please let me know if you think any improvements are required.
ID: 38373 · Report as offensive     Reply Quote
mmonnin

Send message
Joined: 22 Mar 17
Posts: 44
Credit: 3,801,950
RAC: 10
Message 38382 - Posted: 22 Mar 2019, 21:51:53 UTC - in response to Message 38373.  

Well, thanks for the tips but I couldn't get this damn CVMFS installed at all no matter what I tried meanwhile. :-\
I'm too inexperienced with Linux to dig into this even more and don't have the time either so I give up here.
Too bad... I hope a Windows app for this will come out sooner or later, if not then so be it...


If you want to try Ubuntu you can use the instructions I gave to my team. It's worked for me on several 18.04 installs. The instructions given at LHC forums are way overly complicated for no reason when CVFMS can be installed from a repository. It's also much quicker than compiling it.

https://www.overclock.net/forum/365-overclock-net-boinc-team/1721752-projects-month-march-2019-a.html#post27876964


Our instructions for the Theory app point to the official CVMFS documentation. Please let me know if you think any improvements are required.


I was referring to the Native ATLAS instructions, which also need CVFMS, but instruct the user to D/L and make the BOINC, CVFMS and singularity apps from source.
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4840

The official CVFMS instructions are the ones I used at OCN.
ID: 38382 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38390 - Posted: 23 Mar 2019, 12:07:24 UTC - in response to Message 38368.  

...
If you want to try Ubuntu you can use the instructions I gave to my team. It's worked for me on several 18.04 installs. The instructions given at LHC forums are way overly complicated for no reason when CVFMS can be installed from a repository. It's also much quicker than compiling it.

https://www.overclock.net/forum/365-overclock-net-boinc-team/1721752-projects-month-march-2019-a.html#post27876964

Well, I'm a bit wondering about myself but I could get it to work!
I didn't like Ubuntu much after I could not get an earlier version to work under the VirtualBox program a few years ago.
But since I had enough time since yesterday I really gave it a try and installed the 18.04 version.
It is a lot more "user friendly" as the version I tried once so I was quite surprised.
After digging into the stuff I even managed to install BOINC and the CVMFS completely and correct and it runs now since a few hours with correct results.
However, the credits for these Theory Native tasks are a laugh so far:
219772254 109580010 10590318 22 Mar 2019, 20:04:43 UTC 23 Mar 2019, 10:45:24 UTC Fertig und Bestätigt 2,305.59 1,406.26 1.47
219818657 109610809 10590318 23 Mar 2019, 10:07:47 UTC 23 Mar 2019, 11:47:01 UTC Fertig und Bestätigt 3,693.08 2,754.01 2.36
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38390 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38392 - Posted: 23 Mar 2019, 16:27:46 UTC - in response to Message 38390.  

After I optimised some settings for my Ubuntu-VB (had only one core reserved for it the first time) I just found out that this Native app seems to be multithreading.
Three tasks are now running parallel with using 2 cores at the same time, doesn't seem that they use more than 2.
Maybe this will have an impact on the credits, will see.
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38392 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 739
Credit: 6,027,140
RAC: 982
Message 38394 - Posted: 23 Mar 2019, 17:11:09 UTC

It will not give more credits.
In fact the jobs aren't real multi-threaded, but 1 job uses several processes so when there are free cores available the job will run faster.
Especially pythia6 and a bit less pythia8 and herwig will profit from a free core.
With your now 6 core-VM you could try to run 5 jobs at once by setting "Use at most 83.34% of the Cpu's in BOINC Manager.
ID: 38394 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38395 - Posted: 23 Mar 2019, 17:33:44 UTC - in response to Message 38394.  

It will not give more credits.

Hm, bad... If it weren't for the hours (needed for WUProp) I would already stop crunching then, wasn't worth the work I put into this imho. ;-\

With your now 6 core-VM you could try to run 5 jobs at once by setting "Use at most 83.34% of the Cpu's in BOINC Manager.

Hm, why shoud I do that? At least hourwise it doesn't seem to be the same to me, I lose a core with runtime...
However, I tried that for the fun of it, but the effect is not what you thought - instead two tasks with 2 cores are running and the last task is suspended...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38395 · Report as offensive     Reply Quote
computezrmle
Avatar

Send message
Joined: 15 Jun 08
Posts: 1133
Credit: 55,676,700
RAC: 100,593
Message 38396 - Posted: 23 Mar 2019, 17:46:48 UTC - in response to Message 38395.  

It will not give more credits.

Hm, bad...

It's clearly marked as beta app.
Both sides, project team and volunteers, have to learn how to deal with it.

This may be a reason not to give full credits for this app ATM.
Otherwise it may attract crunchers who are only interested in credits instead of science or knowledge.
ID: 38396 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 739
Credit: 6,027,140
RAC: 982
Message 38400 - Posted: 23 Mar 2019, 18:07:04 UTC - in response to Message 38395.  

However, I tried that for the fun of it, but the effect is not what you thought - instead two tasks with 2 cores are running and the last task is suspended...
How did you set Max # of CPUs in your project preferences?
If # is 2 set it to 1 and no limit for # of Jobs. BOINC will pick up those settings after Update and only for new loaded tasks.
ID: 38400 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38401 - Posted: 23 Mar 2019, 19:49:24 UTC - in response to Message 38400.  
Last modified: 23 Mar 2019, 19:55:04 UTC

If # is 2 set it to 1 and no limit for # of Jobs. BOINC will pick up those settings after Update and only for new loaded tasks.

Oh, I see, that was set to "no limits" of course - normally I don't fiddle much on any of these settings if not needed.

Edit:
Hm, nope, doesn't change it, tasks are still running in "2-core"-mode...

Edit 2:
Oh wait, only new loaded tasks? Well, then I have bad luck and have to wait, there are no new ones since the change.
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38401 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38404 - Posted: 24 Mar 2019, 12:33:17 UTC
Last modified: 24 Mar 2019, 12:34:04 UTC

Well, it does work as expected now but there are only two remaining tasks and the manager doesn't get no new work - although server status shows that enough Native tasks are there.
What's up with that?
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38404 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 739
Credit: 6,027,140
RAC: 982
Message 38406 - Posted: 24 Mar 2019, 16:18:17 UTC - in response to Message 38404.  

What's up with that?
The server scheduler is sometimes confused, how to interpret the max # settings.
You could try the use of an app_config.xml in LHC@home project directory.
Running only Theory native it should look like this:

<app_config>
 <project_max_concurrent>7</project_max_concurrent>
 <app>
 <name>TheoryN</name>
  <max_concurrent>7</max_concurrent>
 </app>
 <app_version>
  <app_name>TheoryN</app_name>
  <plan_class>native_theory</plan_class>
  <avg_ncpus>1</avg_ncpus>
 </app_version>
</app_config>

After you have loaded the app_config.xml with the use of BOINC Manager, you could set both Max # to no limit.
ID: 38406 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38407 - Posted: 24 Mar 2019, 17:03:36 UTC - in response to Message 38406.  

The server scheduler is sometimes confused, how to interpret the max # settings.
...

Yeah I already suspected that. As soon as I changed it back enough tasks went in.
Thanks for the app_config but unfortunately when I want to save the file with gedit in the project directory Ubuntu doesn't let me access the 'projects'-folder for some reason...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38407 · Report as offensive     Reply Quote
bronco

Send message
Joined: 13 Apr 18
Posts: 443
Credit: 8,201,701
RAC: 10,611
Message 38411 - Posted: 24 Mar 2019, 19:55:45 UTC - in response to Message 38407.  

The server scheduler is sometimes confused, how to interpret the max # settings.
...

Yeah I already suspected that. As soon as I changed it back enough tasks went in.
Thanks for the app_config but unfortunately when I want to save the file with gedit in the project directory Ubuntu doesn't let me access the 'projects'-folder for some reason...

You need root privilege to create/edit that file otherwise it's read only I usually do sudo nano ../app_config.xml. Nano is a minimal editor but it works well enough for short files.
ID: 38411 · Report as offensive     Reply Quote
mmonnin

Send message
Joined: 22 Mar 17
Posts: 44
Credit: 3,801,950
RAC: 10
Message 38421 - Posted: 24 Mar 2019, 23:50:47 UTC - in response to Message 38407.  

The server scheduler is sometimes confused, how to interpret the max # settings.
...

Yeah I already suspected that. As soon as I changed it back enough tasks went in.
Thanks for the app_config but unfortunately when I want to save the file with gedit in the project directory Ubuntu doesn't let me access the 'projects'-folder for some reason...


To open the file manager with elevated access:

sudo nautilus


Or
sudo gedit /var/lib/boinc-client/lhcathome.cern.ch_lhcathome


I run the Theory Native tasks as single threaded with app_config since the only use about 1.1-1.5 threads but are scheduled by default to use 2.[/code]
ID: 38421 · Report as offensive     Reply Quote
Profile DoctorNow
Avatar

Send message
Joined: 17 Sep 04
Posts: 19
Credit: 308,023
RAC: 40
Message 38425 - Posted: 25 Mar 2019, 5:16:34 UTC

Many thanks for the infos guys. It works now like it should. :-)
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 38425 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 325
Credit: 10,743,552
RAC: 22,423
Message 38426 - Posted: 25 Mar 2019, 8:35:29 UTC - in response to Message 38421.  
Last modified: 25 Mar 2019, 8:43:11 UTC

I run the Theory Native tasks as single threaded with app_config since the only use about 1.1-1.5 threads but are scheduled by default to use 2.

Thanks. I am now running Theory Native on a single thread, and they are doing well. It is more convenient for the BOINC scheduler too.
(It helps to keep a core free, since the Native Theory use more than one, but I leave 2 cores free anyway to support a GPU on Folding, so it works out quite well for CPU efficiency.)
ID: 38426 · Report as offensive     Reply Quote

Message boards : Theory Application : Installation of CVMFS


©2019 CERN