cluster.toml.example

[cluster]
name = "my-cluster"
domain = "example.com"
acme_email = "ops@example.com"
log_level = "info"

# Single-node setup — just omit the [[node]] sections and orca runs everything locally.
# Multi-node — list your nodes below. The first node to run `orca server` becomes the leader.

# [[node]]
# address = "10.0.0.1"
# labels = { zone = "eu-1", role = "general" }
#
# [[node]]
# address = "10.0.0.2"
# labels = { zone = "eu-1", role = "general" }

# GPU node — declare GPUs so the scheduler can place GPU workloads
# [[node]]
# address = "10.0.0.3"
# labels = { zone = "eu-1", role = "gpu" }
#
# [[node.gpus]]
# vendor = "nvidia"
# count = 2
# model = "A100"

# -- AI Operations Assistant --
# Enables: `orca ask`, conversational alerts, smart import analysis, log summarization.
# Works with any OpenAI-compatible API (LiteLLM, Ollama, vLLM, OpenAI, etc.)

# [ai]
# provider = "litellm"
# endpoint = "https://llm.example.com"
# model = "qwen3-30b"
# api_key = "${secrets.ai_api_key}"
#
# [ai.alerts]
# enabled = true
# analysis_interval_secs = 60
#
# [ai.alerts.channels]
# slack = "https://hooks.slack.com/services/..."
# webhook = "https://my-pagerduty-webhook/..."
#
# [ai.auto_remediate]
# restart_crashed = true
# scale_on_pressure = false
# rollback_on_failure = false

# [observability]
# otlp_endpoint = "https://signoz.example.com"
#
# [observability.alerts]
# webhook = "https://hooks.slack.com/..."
# email = "ops@example.com"