Merge pull request #4230 from KCSesh/api-time-slice

Add Nvidia Time-Slicing

Release.toml

@@ -359,4 +359,5 @@ version = "1.25.0"
 ]
 "(1.24.1, 1.25.0)" = [
     "migrate_v1.25.0_kubernetes-service-config.lz4",
+    "migrate_v1.25.0_kubelet-device-plugins-time-slicing-settings.lz4",
 ]

sources/Cargo.toml

@@ -60,6 +60,7 @@ members = [
     "settings-migrations/v1.24.1/aws-control-container-v0-7-16",
     "settings-migrations/v1.24.1/public-control-container-v0-7-16",
     "settings-migrations/v1.25.0/kubernetes-service-config",
+    "settings-migrations/v1.25.0/kubelet-device-plugins-time-slicing-settings",
 
     "settings-plugins/aws-dev",
     "settings-plugins/aws-ecs-1",
@@ -134,13 +135,13 @@ version = "0.1.0"
 
 [workspace.dependencies.bottlerocket-modeled-types]
 git = "https://github.com/bottlerocket-os/bottlerocket-settings-sdk"
-tag = "bottlerocket-settings-models-v0.5.0"
-version = "0.5.0"
+tag = "bottlerocket-settings-models-v0.6.0"
+version = "0.6.0"
 
 [workspace.dependencies.bottlerocket-settings-models]
 git = "https://github.com/bottlerocket-os/bottlerocket-settings-sdk"
-tag = "bottlerocket-settings-models-v0.5.0"
-version = "0.5.0"
+tag = "bottlerocket-settings-models-v0.6.0"
+version = "0.6.0"
 
 [workspace.dependencies.bottlerocket-settings-plugin]
 git = "https://github.com/bottlerocket-os/bottlerocket-settings-sdk"
@@ -149,7 +150,7 @@ version = "0.1.0"
 
 [workspace.dependencies.settings-extension-oci-defaults]
 git = "https://github.com/bottlerocket-os/bottlerocket-settings-sdk"
-tag = "bottlerocket-settings-models-v0.5.0"
+tag = "bottlerocket-settings-models-v0.6.0"
 version = "0.1.0"
 
 [profile.release]

sources/settings-migrations/v1.25.0/kubelet-device-plugins-time-slicing-settings/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "kubelet-device-plugins-time-slicing-settings"
+version = "0.1.0"
+authors = ["Kyle Sessions <[email protected]>"]
+license = "Apache-2.0 OR MIT"
+edition = "2021"
+publish = false
+# Don't rebuild crate just because of changes to README.
+exclude = ["README.md"]
+
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+migration-helpers.workspace = true

sources/settings-migrations/v1.25.0/kubelet-device-plugins-time-slicing-settings/src/main.rs

@@ -0,0 +1,23 @@
+use migration_helpers::common_migrations::AddSettingsMigration;
+use migration_helpers::{migrate, Result};
+use std::process;
+
+/// We added new settings for configuring the NVIDIA k8s device plugin.
+fn run() -> Result<()> {
+    migrate(AddSettingsMigration(&[
+        "settings.kubelet-device-plugins.nvidia.device-sharing-strategy",
+        "settings.kubelet-device-plugins.nvidia.time-slicing.replicas",
+        "settings.kubelet-device-plugins.nvidia.time-slicing.rename-by-default",
+        "settings.kubelet-device-plugins.nvidia.time-slicing.fail-requests-greater-than-one",
+    ]))
+}
+
+// Returning a Result from main makes it print a Debug representation of the error, but with Snafu
+// we have nice Display representations of the error, so we wrap "main" (run) and print any error.
+// https://github.com/shepmaster/snafu/issues/110
+fn main() {
+    if let Err(e) = run() {
+        eprintln!("{}", e);
+        process::exit(1);
+    }
+}

sources/shared-defaults/nvidia-k8s-device-plugin.toml

@@ -21,3 +21,4 @@ affected-services = ["nvidia-k8s-device-plugin"]
 pass-device-specs = true
 device-id-strategy="index"
 device-list-strategy="volume-mounts"
+device-sharing-strategy="none"