Ray exposes configuration throughDocumentation Index
Fetch the complete documentation index at: https://ray-preview.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
ray.init parameters, environment variables, and command-line flags to ray start.
Initialize the runtime
| Parameter | What it does |
|---|---|
address | Connect to an existing cluster. Use "auto" for the local Ray service or a head-node address. |
namespace | Logical grouping for named actors. |
runtime_env | Default runtime environment for the job. |
num_cpus, num_gpus | Cap resources used by the local Ray instance. |
object_store_memory | Object store size in bytes. |
_temp_dir | Where to write logs, sockets, and spilled objects. |
Start a head node
Start a worker node
Environment variables
| Variable | Effect |
|---|---|
RAY_ADDRESS | Default address for ray.init(). |
RAY_RUNTIME_ENV_TIMEOUT_SECONDS | Maximum time to install runtime env. |
RAY_DEDUP_LOGS | Deduplicate identical log lines from many workers. |
RAY_TMPDIR | Override the temp directory. |
RAY_BACKEND_LOG_LEVEL | Log level for Ray’s C++ core. |
Object store configuration
Logging
Ray writes per-worker logs into/tmp/ray/session_*/logs. To control log level:
RAY_ROTATION_MAX_BYTES and RAY_ROTATION_BACKUP_COUNT.
Resource labels
Custom resources let you pin tasks and actors to specific node classes.Next steps
Scheduling
Resource requests, placement groups, and locality.
Troubleshooting
Diagnosing common Ray issues.