[SLURM] Outage Thursday May 3 | Adding GPU servers!!!

[Phil Kauffman via Slurm]

I will be taking the cluster down tomorrow to add a gpu server. The 
methods for sharing GPU resources on one machine is currently untested 
around here so please forgive any extended outage. Worst case, the 
cluster will be in a working state over the weekend.

FWIW, I won't stop you from submitting jobs during this outage. Just 
know that at anytime tomorrow I may restart the cluster nodes or slurmd 
service which would cancel your job. So feel free to play Russian 
roulette with your jobs. :P

I'll let you know when I am done and will include some documentation on 
how to use the GPU server.

-- 
Phil Kauffman
Systems Administrator
Dept. of Computer Science
University of Chicago
kauffman at cs.uchicago.edu
773-702-3913