Configuration

The agent reads configuration from a YAML file at /etc/nebula/config.yaml. Every setting can also be overridden with environment variables.

Config File

server:
  api_key: "your-api-key-here"
  endpoint: "https://api.sysmos.org"
  grpc_endpoint: "grpc.sysmos.org:443"

collection:
  interval: 10s         # How often to collect metrics
  ebpf_enabled: true    # Enable eBPF-based metrics
  processes: true       # Collect per-process metrics
  network: true         # Collect network metrics
  disk: true            # Collect disk I/O metrics

buffer:
  enabled: true
  max_size: 500MB       # Local buffer size for offline resilience
  path: /var/lib/sysmos/buffer

logging:
  level: info           # debug, info, warn, error
  file: /var/log/sysmos/agent.log

Environment Variable Overrides

Any config key can be overridden using environment variables with the pattern NEBULA_<SECTION>__<KEY>:

# Override the API key
export NEBULA_SERVER__API_KEY="your-key"

# Change collection interval
export NEBULA_COLLECTION__INTERVAL="30s"

# Disable eBPF metrics
export NEBULA_COLLECTION__EBPF_ENABLED="false"

# Set log level to debug
export NEBULA_LOGGING__LEVEL="debug"

Note the double underscore (__) between section and key.

Key Settings

Collection Interval

Controls how frequently metrics are gathered and sent. Lower intervals give more granularity but increase CPU and network usage.

Buffer Settings

The local buffer (backed by BadgerDB) stores metrics when the network is unavailable:

• max_size — Maximum disk space for the buffer (default: 500 MB)
• When the buffer fills up, oldest metrics are dropped to make room
• On reconnection, buffered data is flushed in chronological order

eBPF Settings

eBPF probes require kernel 5.8+ with BTF. If your kernel doesn't support it, set ebpf_enabled: false to fall back to standard /proc and /sys based collection.

Apply Configuration Changes

After editing the config file, restart the agent:

sudo systemctl restart sysmos-agent