Run a single node
Start a local coordinator, executor, status API, and Flight SQL endpoint.
Single-node mode exercises service and transport boundaries on one machine. It is useful for integration testing before evaluating the distributed Preview.
Build the CLI
cargo build -p krishivStart local services
cargo run -p krishiv -- local start \
--data-dir ./.krishiv-single-node \
--http-addr 127.0.0.1:2002The status UI and HTTP health surface listen on the configured address. The local command manages the coordinator and executor processes together. It also prints separate gRPC, HTTP, and Flight URLs; Flight uses the first free port at or above 2003.
Check status
cargo run -p krishiv -- local status \
--data-dir ./.krishiv-single-node
curl http://127.0.0.1:2002/healthz
curl http://127.0.0.1:2002/readyzPoint clients at Flight SQL
export KRISHIV_COORDINATOR_URL=http://127.0.0.1:2003
cargo run -p krishiv -- sql \
--mode single-node \
--query "SELECT 42 AS answer"Replace the example value with the exact Flight: URL printed by local start
or local status; do not assume that port 2003 was free.
Stop the node
cargo run -p krishiv -- local stop \
--data-dir ./.krishiv-single-nodeEvaluation topology
Single-node mode is not evidence of high availability. Back up any durable metadata/state paths and test restore behavior before retaining important data.