sPHENIX discussion of software

[Chris Pinkenburg <pinkenburg AT bnl.gov>]

Hi folks,

we got another 1000 condor slots last week, we are at a total of 16000 
slots now. We have no restrictions on the number of jobs a user can run.

These pretty awesome computing resources you have at your finger tips 
come with real responsibilities on your side. A continuing issue has 
been excessive amount of (basically useless) text written to the condor 
log files. This severely impacts the performance of our central storage. 
To alleviate this - the condor logs (output, log, error) when written to 
gpfs (working group disks, scratch, sphenix filesystems) go to a 
separate SSD based storage WITH VERY LIMITED CAPACITY. Someone submitted 
sphenix simulations at 2am which created a >3GB log for each job 
resulting in 8TB of useless stuff in less than one hour. This filled up 
the ssd storage which in turn caused work for the people managing our 
storage (in the future logs from this user will go again to SATA storage 
which will just slow those jobs down when writing excessive logs).

So please - keep an eye on what you are doing and be mindful about the 
impact if you multiply whatever you are doing by 10000.

Chris

--

*********************************************************

Christopher H. Pinkenburg      ; pinkenburg AT bnl.gov

Brookhaven National Laboratory ; phone: (631) 344-5692
Physics Department Bldg 510 C  ; fax:   (631) 344-3253
Upton, NY 11973-5000

http://www.phenix.bnl.gov/~pinkenbu/

*********************************************************