hiro/ray
1
0
Fork 0
mirror of https://github.com/vale981/ray synced 2025-03-05 10:01:43 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

use smaller instance for scheduling tests (#25635)

Browse source 
m5.16xlarge instances have 64 CPU and 256GB memory, which are overkill for scheduling tests that do not have a lot of computations. Use smaller instance m5.4xlarge to save cost and make allocating instances easier.
This commit is contained in:
mwtian 2022-06-10 10:09:35 -07:00 • committed by GitHub
parent 271c7d73ac
commit 1483c4553c
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 5 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
release/benchmarks/scheduling.yaml
View file

@ -5,15 +5,18 @@ max_workers: 999
head_node_type:
name: head_node
instance_type: m5.16xlarge
instance_type: m5.4xlarge
resources:
# Assume the node has 64 CPU instead of 16.
# This should be fine since each task has little
# computation in scheduling tests.
cpu: 64
custom_resources:
node: 1
worker_node_types:
- name: worker_node
instance_type: m5.16xlarge
instance_type: m5.4xlarge
min_workers: 31
max_workers: 31
use_spot: false