Thank you for the quick response. See below for my reply. On 1/4/16, 6:25 PM, "[email protected]" <[email protected]> wrote:
> >Quoting "GOLPAYEGANI, NAVID (GSFC-6190)" <[email protected]>: >> Hi, >> SLURM newbie here. Anybody have suggestions on how to do the >>scheduling >> for a SLURM cluster for web submissions? We take orders from a webpage >>and >> submit them as a single account > >By "account" do you mean Slurm "bank account" (and different user IDs)? Yes a single SLURM bank account which is the same as a single linux account in our case. I don't think we can realistically create individual Slurm bank accounts for all the web users since the system is open to the world. Anybody can request our data products with some sort of post processing and all we ask for is for their email address (honor system we don't verify). The purpose of the email address is to notify them when their post processing job has finished and is ready for download. > >There are a bunch of limits that can be configured by user, account, >QOS (quality of service). See: >http://slurm.schedmd.com/resource_limits.html > >Job's can also be assigned a priority to that under-serviced users and >accounts get higher job priorities (i.e. jobs may not prioritized by >submit order). For more information about job priorities, see: >http://slurm.schedmd.com/priority_multifactor.html I've read those but those would require each web user to be listed with an account in our Slurm database, correct? My initial instinct is that that's a bad for several reasons. It'd be ever growing list of users, it'd initially start off with 10s of thousands of entries, and assuming Slurm bank accounts are alphanumeric only we'd have to convert email addresses to unique alphanumeric account string (md5?). Am I on the right path with my thinking? Thanks again for the quick response and all the help so far. Navid
