-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Paul,
On 16/01/14 03:29, Paul Edmon wrote: > Thoughts? If this doesn't exist it may be a good thing to add into > the next version of SLURM as it would be handy in our environment. Whilst we don't use fairshare (we have allocated quota limits for projects) we have the same issue with memory use that you do. We worked our way around it (originally with Torque+Moab+Gold, but now with Slurm) by instituting a nightly surcharge for jobs that finished that day that used more than 16GB/core. We were lucky in that we already had a system for managing users & projects that includes usage and so it was already in a database and so pretty easy to implement (surcharges are also recorded in the database so our command line tools for users can report what's going on). The major issue we've found is that currently we only have the one pot to dip into (CPU hours) and so if a project requests an awful lot of memory then there is the potential for them to swamp the machine with jobs. For that reason we are considering having a separate pot for memory and hit that until we exhaust it and then hit CPU hours. Best of luck! Chris - -- Christopher Samuel Senior Systems Administrator VLSCI - Victorian Life Sciences Computation Initiative Email: sam...@unimelb.edu.au Phone: +61 (0)3 903 55545 http://www.vlsci.org.au/ http://twitter.com/vlsci -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.14 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlLXELkACgkQO2KABBYQAh8rewCfXU/JQ3HNnI/O6cNOwic7ZU8B UzoAn3pK6eDzMQYrNHK1fdLto6mMOwaM =IDzj -----END PGP SIGNATURE-----
