Skip to content

Commit

Permalink
Merge pull request #4940 from consideRatio/pr/adjust-victor
Browse files Browse the repository at this point in the history 
victor: transition to node sharing and adjust server options for that
  • Loading branch information
@consideRatio
consideRatio authored Oct 4, 2024
2 parents 3a0b168 + a969bb6 commit a734b7c
Show file tree
Hide file tree
Showing 4 changed files with 181 additions and 114 deletions.
67 changes: 0 additions & 67 deletions config/clusters/victor/common.values.yaml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -52,73 +52,6 @@ basehub:
- einatlev-ldeo
- SamKrasnoff
singleuser:
profileList:
# The mem-guarantees are here so k8s doesn't schedule other pods
# on these nodes.
- display_name: "Small: m5.large"
description: "~2 CPU, ~8G RAM"
default: true
kubespawner_override:
# Explicitly unset mem_limit, so it overrides the default memory limit we set in
# basehub/values.yaml
mem_limit: 8G
mem_guarantee: 6.5G
node_selector:
node.kubernetes.io/instance-type: m5.large
profile_options: &profile_options
image:
display_name: Image
choices:
a-victor-notebook:
display_name: Victor Notebook
default: true
kubespawner_override:
image: quay.io/volcanocyber/victor-notebook:a045ad3616d1
b-pytorch-notebook:
display_name: Pangeo ML Notebook (Pytorch)
kubespawner_override:
image: "quay.io/pangeo/pytorch-notebook:2024.09.11"
c-ml-notebook:
display_name: Pangeo ML Notebook (Tensorflow)
kubespawner_override:
image: "quay.io/pangeo/ml-notebook:2024.09.11"
- display_name: "Medium: m5.xlarge"
description: "~4 CPU, ~15G RAM"
kubespawner_override:
mem_limit: 15G
mem_guarantee: 12G
node_selector:
node.kubernetes.io/instance-type: m5.xlarge
profile_options: *profile_options
- display_name: "Large: m5.2xlarge"
description: "~8 CPU, ~30G RAM"
kubespawner_override:
mem_limit: 30G
mem_guarantee: 25G
node_selector:
node.kubernetes.io/instance-type: m5.2xlarge
profile_options: *profile_options
- display_name: "Huge: m5.8xlarge"
description: "~16 CPU, ~60G RAM"
kubespawner_override:
mem_limit: 60G
mem_guarantee: 50G
node_selector:
node.kubernetes.io/instance-type: m5.8xlarge
profile_options: *profile_options
- display_name: NVIDIA Tesla T4, ~16 GB, ~4 CPUs
description: "Start a container on a dedicated node with a GPU"
slug: "gpu"
kubespawner_override:
environment:
NVIDIA_DRIVER_CAPABILITIES: compute,utility
mem_limit: null
mem_guarantee: 14G
node_selector:
node.kubernetes.io/instance-type: g4dn.xlarge
extra_resource_limits:
nvidia.com/gpu: "1"
profile_options: *profile_options
defaultUrl: /lab
scheduling:
userScheduler:
Expand Down
106 changes: 106 additions & 0 deletions config/clusters/victor/prod.values.yaml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,3 +13,109 @@ basehub:
config:
GitHubOAuthenticator:
oauth_callback_url: https://victor.2i2c.cloud/hub/oauth_callback
singleuser:
profileList:
# IMPORTANT: Staging and prod's profileList's are meant to be kept
# equivalent with the exception that staging adds
# unlisted_choice to pick a custom image. If you update
# either, update the other as well.
#
- display_name: CPU Only
profile_options: &profile_options
image: &profile_option_image
display_name: Image
choices:
a-victor-notebook:
display_name: Victor Notebook
default: true
kubespawner_override:
image: quay.io/volcanocyber/victor-notebook:a045ad3616d1
b-pytorch-notebook:
display_name: Pangeo ML Notebook (Pytorch)
kubespawner_override:
image: "quay.io/pangeo/pytorch-notebook:2024.09.11"
c-ml-notebook:
display_name: Pangeo ML Notebook (Tensorflow)
kubespawner_override:
image: "quay.io/pangeo/ml-notebook:2024.09.11"
resource_allocation: &profile_option_resource_allocation
display_name: Resource Allocation
choices:
mem_1_9:
default: true
display_name: 1.9 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 1991244775
mem_limit: 1991244775
cpu_guarantee: 0.2328125
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_3_7:
display_name: 3.7 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 3982489550
mem_limit: 3982489550
cpu_guarantee: 0.465625
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_7_4:
display_name: 7.4 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 7964979101
mem_limit: 7964979101
cpu_guarantee: 0.93125
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_14_8:
display_name: 14.8 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 15929958203
mem_limit: 15929958203
cpu_guarantee: 1.8625
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_29_7:
display_name: 29.7 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 31859916406
mem_limit: 31859916406
cpu_guarantee: 3.725
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_60_6:
display_name: 60.6 GB RAM, upto 15.6 CPUs
kubespawner_override:
mem_guarantee: 65094448840
mem_limit: 65094448840
cpu_guarantee: 7.8475
cpu_limit: 15.695
node_selector:
node.kubernetes.io/instance-type: r5.4xlarge
mem_121_2:
display_name: 121.2 GB RAM, upto 15.6 CPUs
kubespawner_override:
mem_guarantee: 130188897681
mem_limit: 130188897681
cpu_guarantee: 15.695
cpu_limit: 15.695
node_selector:
node.kubernetes.io/instance-type: r5.4xlarge
- display_name: NVIDIA Tesla T4, ~16 GB, ~4 CPUs
description: "Start a container on a dedicated node with a GPU"
slug: "gpu"
kubespawner_override:
environment:
NVIDIA_DRIVER_CAPABILITIES: compute,utility
mem_limit: null
mem_guarantee: 14G
node_selector:
node.kubernetes.io/instance-type: g4dn.xlarge
extra_resource_limits:
nvidia.com/gpu: "1"
profile_options:
image: *profile_option_image
118 changes: 75 additions & 43 deletions config/clusters/victor/staging.values.yaml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -15,26 +15,14 @@ basehub:
oauth_callback_url: https://staging.victor.2i2c.cloud/hub/oauth_callback
singleuser:
profileList:
#=== Below are copied from common file ===#
# IMPORTANT: Staging and prod's profileList's are meant to be kept
# equivalent with the exception that staging adds
# unlisted_choice to pick a custom image. If you update
# either, update the other as well.
#
# But, they have been adjusted to include unlisted_choice to pick a
# custom image.
#

# The mem-guarantees are here so k8s doesn't schedule other pods
# on these nodes.
- display_name: "Small: m5.large"
description: "~2 CPU, ~8G RAM"
default: true
kubespawner_override:
# Explicitly unset mem_limit, so it overrides the default memory limit we set in
# basehub/values.yaml
mem_limit: 8G
mem_guarantee: 6.5G
node_selector:
node.kubernetes.io/instance-type: m5.large
- display_name: CPU Only
profile_options: &profile_options
image:
image: &profile_option_image
display_name: Image
choices:
a-victor-notebook:
Expand All @@ -57,30 +45,73 @@ basehub:
validation_message: "Must be a publicly available docker image, of form <image-name>:<tag>"
kubespawner_override:
image: "{value}"
- display_name: "Medium: m5.xlarge"
description: "~4 CPU, ~15G RAM"
kubespawner_override:
mem_limit: 15G
mem_guarantee: 12G
node_selector:
node.kubernetes.io/instance-type: m5.xlarge
profile_options: *profile_options
- display_name: "Large: m5.2xlarge"
description: "~8 CPU, ~30G RAM"
kubespawner_override:
mem_limit: 30G
mem_guarantee: 25G
node_selector:
node.kubernetes.io/instance-type: m5.2xlarge
profile_options: *profile_options
- display_name: "Huge: m5.8xlarge"
description: "~16 CPU, ~60G RAM"
kubespawner_override:
mem_limit: 60G
mem_guarantee: 50G
node_selector:
node.kubernetes.io/instance-type: m5.8xlarge
profile_options: *profile_options
resource_allocation: &profile_option_resource_allocation
display_name: Resource Allocation
choices:
mem_1_9:
default: true
display_name: 1.9 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 1991244775
mem_limit: 1991244775
cpu_guarantee: 0.2328125
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_3_7:
display_name: 3.7 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 3982489550
mem_limit: 3982489550
cpu_guarantee: 0.465625
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_7_4:
display_name: 7.4 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 7964979101
mem_limit: 7964979101
cpu_guarantee: 0.93125
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_14_8:
display_name: 14.8 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 15929958203
mem_limit: 15929958203
cpu_guarantee: 1.8625
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_29_7:
display_name: 29.7 GB RAM, upto 3.7 CPUs
kubespawner_override:
mem_guarantee: 31859916406
mem_limit: 31859916406
cpu_guarantee: 3.725
cpu_limit: 3.725
node_selector:
node.kubernetes.io/instance-type: r5.xlarge
mem_60_6:
display_name: 60.6 GB RAM, upto 15.6 CPUs
kubespawner_override:
mem_guarantee: 65094448840
mem_limit: 65094448840
cpu_guarantee: 7.8475
cpu_limit: 15.695
node_selector:
node.kubernetes.io/instance-type: r5.4xlarge
mem_121_2:
display_name: 121.2 GB RAM, upto 15.6 CPUs
kubespawner_override:
mem_guarantee: 130188897681
mem_limit: 130188897681
cpu_guarantee: 15.695
cpu_limit: 15.695
node_selector:
node.kubernetes.io/instance-type: r5.4xlarge
- display_name: NVIDIA Tesla T4, ~16 GB, ~4 CPUs
description: "Start a container on a dedicated node with a GPU"
slug: "gpu"
Expand All @@ -93,4 +124,5 @@ basehub:
node.kubernetes.io/instance-type: g4dn.xlarge
extra_resource_limits:
nvidia.com/gpu: "1"
profile_options: *profile_options
profile_options:
image: *profile_option_image
4 changes: 0 additions & 4 deletions eksctl/victor.jsonnet
View file
Original file line number Diff line number Diff line change
Expand Up @@ -25,10 +25,6 @@ local nodeAz = "us-west-2a";
// A `node.kubernetes.io/instance-type label is added, so pods
// can request a particular kind of node with a nodeSelector
local notebookNodes = [
{ instanceType: "m5.large" },
{ instanceType: "m5.xlarge" },
{ instanceType: "m5.2xlarge" },
{ instanceType: "m5.8xlarge" },
{ instanceType: "r5.xlarge" },
{ instanceType: "r5.4xlarge" },
{ instanceType: "r5.16xlarge" },
Expand Down

0 comments on commit a734b7c

Please sign in to comment.