Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 9 additions & 8 deletions chart/values.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -86,14 +86,15 @@ admission:
# GPU Preemption Configuration
# Enables utilization-based preemption of idle GPU workloads
gpuPreemption:
enabled: false
metricsEndpoint: "" # AMD GPU metrics exporter URL (required when enabled)
pollingInterval: "15s" # How often the scraper polls the metrics endpoint
defaultThreshold: "5" # Default utilization threshold (%) below which a workload is idle
defaultGracePeriod: "10m" # Default idle duration before preemption eligibility
defaultPolicy: "OnPressure" # Default preemption policy (OnPressure or Always)
defaultAggregation: "Max" # Default multi-pod aggregation (Min, Max, or Avg)
defaultTTL: "24h" # Default TTL for terminal GpuWorkload CRs ("0" = retain forever)
enabled: true
# Replace with the actual in-cluster URL of your AMD GPU metrics exporter.
metricsEndpoint: "http://lgtm-stack.otel-lgtm-stack.svc.cluster.local:9090/federate?match[]=gpu_gfx_activity"
pollingInterval: "5s"
defaultThreshold: "5"
defaultGracePeriod: "10m"
defaultPolicy: "OnPressure"
defaultAggregation: "Max"
defaultTTL: "24h"

# Optional KaiwoConfig - global operator configuration
# If provided, a KaiwoConfig resource will be created
Expand Down
Loading