Message boards : News : New Release (v44)
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Laurence
Project administrator
Project developer
Project tester
Avatar

Send message
Joined: 12 Sep 14
Posts: 1069
Credit: 334,882
RAC: 0
Message 53 - Posted: 12 Mar 2015, 9:13:46 UTC

A new app version has been release (v44). The main fixed has been to an issue whereby jobs were failing on machines due to configuration files in CVMFS not being discovered automatically. Various other minor things have been clean up. The image is now small (15MB) but will download about 1GB of files via CVMFS. This may be changed in a future release where we will increase the size of the download image by already including may of the needed files.

Please let us know how things are going by posting on the message boards.

Thanks,

Laurence
ID: 53 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Crystal Pellet
Volunteer tester

Send message
Joined: 13 Feb 15
Posts: 1188
Credit: 861,475
RAC: 2
Message 54 - Posted: 12 Mar 2015, 9:51:41 UTC - in response to Message 53.  

To add the remark, that cause CMS is now running fine within the VM, no need for the reduced runtime for the BOINC-task.

With this version the runtime for a BOINC-task is extended to 24 hours.
ID: 54 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Crystal Pellet
Volunteer tester

Send message
Joined: 13 Feb 15
Posts: 1188
Credit: 861,475
RAC: 2
Message 55 - Posted: 12 Mar 2015, 10:02:04 UTC - in response to Message 54.  

Question:

Could you please explain why there is an idle time of about 15-20 minutes between 2 successive jobs within the VM?
ID: 55 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ivan
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar

Send message
Joined: 20 Jan 15
Posts: 1139
Credit: 8,310,612
RAC: 728
Message 56 - Posted: 12 Mar 2015, 11:21:22 UTC - in response to Message 55.  

Question:

Could you please explain why there is an idle time of about 15-20 minutes between 2 successive jobs within the VM?

There's a lot of set-up to be done -- for example the configuration database has to be interrogated and the detector geometry built, so a lot of files have to be read. Also since this is simulation GEANT too has to read in parameter files, etc. If you do a 'strace -p' on the cmsRun PID you can see some of the filenames flashing by. You might also notice that the first event takes longer than the rest; this is because much of the set-up is "lazy" -- it's not done until the first time an object is called (i.e. the object instantiation).
That said, I'm not sure why there is a pause after the job finishes, something in the clean-up code I guess.
Now that we have solved the SITECONF problem, we should increase the number of events per job to increase the duty cycle, although we have to watch the size of the output files. Currently the five events, which are the heart of the simulation, take three minutes, including the first-event set-up time, so efficiency is not high just yet.
ID: 56 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : News : New Release (v44)


©2024 CERN