Bug #12908

install docs for slurm don't allow multiple jobs per node

Added by Joshua Randall about 2 years ago. Updated about 2 years ago.

Status:
Resolved
Priority:
Normal
Assigned To:
-
Category:
Documentation
Target version:
-
Start date:
Due date:
% Done:

100%

Estimated time:
Story points:
-

Description

The install docs for setting up SLURM (http://doc.arvados.org/install/crunch2-slurm/install-slurm.html) include a configuration which disables SLURM's consumable resources support and forces assigning jobs to whole nodes:

SelectType=select/linear

There should at least be a mention in the docs that if you want multiple jobs per node (i.e. what I'd assume is the default desire for Crunch v2), you need something more like:

SelectType=select/cons_res
SelectTypeParameters=CR_CPU_Memory

Issue 6520 (https://dev.arvados.org/issues/6520) has that configuration along with guidance on how to configure SLURM for node management, but this also appears to not be mentioned in the documentation:

SelectType=select/cons_res
SelectTypeParameters=CR_CPU_Memory
SuspendTime=600
ResumeProgram=/usr/bin/create-node
SuspendProgram=/usr/bin/destroy-node

I suggest either changing the default configuration to use consumable resources or else adding it as an option in the install docs.


Related issues

Related to Arvados - Bug #12199: Don't schedule jobs on nodes which are too much bigger than requestedResolved01/29/2018

Associated revisions

Revision cf8c0978
Added by Tom Clegg about 2 years ago

Merge branch '12908-slurm-cons-res'

fixes #12908

Arvados-DCO-1.1-Signed-off-by: Tom Clegg <>

History

#1 Updated by Tom Clegg about 2 years ago

  • Related to Bug #12199: Don't schedule jobs on nodes which are too much bigger than requested added

#2 Updated by Anonymous about 2 years ago

  • Status changed from New to Resolved
  • % Done changed from 0 to 100

Also available in: Atom PDF