Bug #13570
closed
[Documentation] add how much extra overhead is added for RAM requirement when spinning up a compute node
Added by Bryan Cosca almost 6 years ago.
Updated over 5 years ago.
Release relationship:
Auto
What happens is c-d-slurm expects nodes to have 95% of the advertised memory size actually usable.
So the number that is specified in ramMin is multiplied by 100/95, which is the actual RAM requirement used to call a compute node.
- Target version set to 2018-06-20 Sprint
- Subject changed from [Documentation] add how much extra overhead is added when spinning up a compute node to [Documentation] add how much extra overhead is added for RAM requirement when spinning up a compute node
- Assigned To set to Tom Morris
- Target version changed from 2018-06-20 Sprint to 2018-07-03 Sprint
- Target version changed from 2018-07-03 Sprint to 2018-07-18 Sprint
- Target version changed from 2018-07-18 Sprint to 2018-08-01 Sprint
- Assigned To changed from Tom Morris to Peter Amstutz
- Status changed from New to In Progress
doc/api/execution
- Always capitalize Keep (I'm unsure about "container." Is the convention to capitalize it or not?)
- Isn't cgroups used in the cloud as well? May not be intended, but sentence ordering kind of makes it sound like cgroups is HPC only
doc/user/cwl/cwl-style
- "Available compute nodes" => "Available instance types" or "Available compute node types"
- Status changed from In Progress to Resolved
Also available in: Atom
PDF
Updated by Bryan Cosca 
Updated by Tom Morris 
Updated by