Message boards : Number crunching : Local control of which subprojects run`2
Message board moderation

To post messages, you must log in.

AuthorMessage
pls

Send message
Joined: 22 Oct 07
Posts: 25
Credit: 808,821
RAC: 0
Message 37272 - Posted: 7 Nov 2018, 21:26:57 UTC

I'm starting a new thread because the existing one is has become pretty muddled.

I want a way to locally control which LHC subprojects will be run. I cna't use the account control because I'm computing through gridcoin and have to run under the gridcoin poll account. This is not a request for gridcoind support.

I wanted to eliminate LHCb tasks, since a the moment they just sit there occupying resources without doing much useful. So added the following to my LHC app_config file

<app>
<name>LHCb</name>
<max_concurrent>0</max_concurrent>
</app>

This isn't working, I'm still getting LHCb tasks..

I know that the app_config is being read and that other items in the app_config are taking effect. Did I miss something I need here?

Thanks,
++PLS
ID: 37272 · Report as offensive     Reply Quote
Gunde

Send message
Joined: 9 Jan 15
Posts: 37
Credit: 272,756,694
RAC: 529,651
Message 37274 - Posted: 7 Nov 2018, 22:53:53 UTC

max_concurrent need to be 1 or higher. To use 0 would not work.
ID: 37274 · Report as offensive     Reply Quote
pls

Send message
Joined: 22 Oct 07
Posts: 25
Credit: 808,821
RAC: 0
Message 37275 - Posted: 8 Nov 2018, 1:07:45 UTC - in response to Message 37274.  

And so it didn't. Is there a way with app_config to have LHCb not run at all?
ID: 37275 · Report as offensive     Reply Quote
captainjack

Send message
Joined: 21 Jun 10
Posts: 25
Credit: 2,936,731
RAC: 4,807
Message 37276 - Posted: 8 Nov 2018, 3:10:56 UTC

pls,

The app_config.xml file does not control which tasks get downloaded. All the app_config.xml file does is control the manner in which the downloaded tasks run (number concurrent, memory size, etc.). The only way to control which tasks get downloaded is by using the project preferences.

A complete description of options for the app_config.xml file can be found at the bottom of https://boinc.berkeley.edu/wiki/client_configuration

Hope that helps.
ID: 37276 · Report as offensive     Reply Quote
pls

Send message
Joined: 22 Oct 07
Posts: 25
Credit: 808,821
RAC: 0
Message 37279 - Posted: 8 Nov 2018, 9:06:51 UTC - in response to Message 37276.  

Yeah, I have that. But while that page gives the format of the options, it doesn't really explain what they DO.

Really, I have this problem because the control that LHC provides over which subproject are run is at the wrong level. It should be at the individual computer level. Not that it can't also be at the account level, serving as the default for individual computers. But needs to be at the computer level, too.

Thanks for the information.
ID: 37279 · Report as offensive     Reply Quote
pls

Send message
Joined: 22 Oct 07
Posts: 25
Credit: 808,821
RAC: 0
Message 37304 - Posted: 11 Nov 2018, 11:10:59 UTC - in response to Message 37279.  

I seem to accidently have solved my problem. I set

<max_concurrent>1</max_concurrent>

and

<avg_ncpus>1.0</avg_ncpus>

I figured that if I have to run LHCb, they can only have 1 cpu.

And I haven't seen an LHCb job since.

Is LHCb not handing out 1 cpu jobs any more?
ID: 37304 · Report as offensive     Reply Quote
computezrmle
Avatar

Send message
Joined: 15 Jun 08
Posts: 1133
Credit: 55,676,450
RAC: 100,636
Message 37305 - Posted: 11 Nov 2018, 12:01:47 UTC - in response to Message 37304.  

... And I haven't seen an LHCb job since.

Is LHCb not handing out 1 cpu jobs any more?

This links/comments will tell you what's going on:
https://lhcathome.cern.ch/lhcathome/server_status.php
https://lhcathome.cern.ch/lhcathome/lhcb_job.php
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4871&postid=37283
ID: 37305 · Report as offensive     Reply Quote
marmot
Avatar

Send message
Joined: 5 Nov 15
Posts: 127
Credit: 6,204,427
RAC: 377
Message 37597 - Posted: 13 Dec 2018, 5:30:08 UTC - in response to Message 37304.  



<avg_ncpus>1.0</avg_ncpus>


I went through the same discussion a year orso back. Not in the pool but can empathize with your dilemma.

It's untested but if you set <avg_ncpus></avg_ncpus> to more cores than your BOINC client has available (or greater than 16 cores), LHC@home server should stop sending WU's which have impossible requirements to fulfill.
ID: 37597 · Report as offensive     Reply Quote

Message boards : Number crunching : Local control of which subprojects run`2


©2019 CERN