Please note: You will need a (user) certificate to access some of the links.
Support for the NGS at Imperial College is provided by the UKI-LT2-IC-HEP site via the ngs.ac.uk VO. To learn about joining the NGS VO, please click here.
The site currently (10/2011) provides approximately 2000 jobs slots using CentOS 5 as the operating system and gLite 3.2 middleware.
The cluster is served by 4 CEs (2 lcgCEs, 2 creamCEs). The nominal share for the NGS is 1%, but no cap is applied, so there is potential to run many more jobs if the cluster is not busy. Users' home directories are shared across the cluster and should not be used to store large files, this should be done in $TMPDIR (= the directory the batch job runs in). Results should be transferred back to the submitters site before the job ends. Some limited storage is available on the Storage Element (SE) attached to the cluster, please contact the site admins if you would like to use this facility.
To direct jobs to the site, please specify
Requirements = RegExp ("ic.ac.uk",other.GlueCEUniqueId);
in your jdl file. It is discouraged to submit to a queue directly, but if you need to see which queues are open to ngs.ac.uk, you can query the site bdii using
ldapsearch -x -H ldap://bdii.grid.hep.ph.ic.ac.uk:2170 -b mds-vo-name=UKI-LT2-IC-HEP,o=grid | perl -00pe 's/\r*\n //g' | grep ngs | grep GlueCEUniqueID
The queues are redundant and feed into the same queue in the underlying batch system.
The maximum job length (CPU *and* Walltime) is 48 h, though we recommend a job length of around 12 h and a maximum memory usage of 2 GB.
For any queries regarding the NGS at Imperial College London please email the site admins or use the GGUS ticketing system, specifying UKI-LT2-IC-HEP as the affected site.
