pub fn shutdown() -> Result<()>Expand description
Stop the engine, joining every thread chDB started.
Call this before a host teardown sequence of its own — global destructors, a finalizing language runtime, a sanitizer exit handler — which would otherwise race the engine’s still-running threads.
A plain exit needs care too. Exiting with a connection still open leaves the
engine to be torn down from inside exit, and if MergeTree merges are
running in the background at that point the process can crash on its way
out. Drop every connection before the process exits, and call this after the
last one is gone: it returns only once no chDB thread is left. Two common
ways a Rust program keeps a connection open at exit: std::process::exit
runs no destructors, and a connection held in a static is never dropped.
This is one-way. Once it succeeds, no further connection can be opened in this process. Calling it again is harmless.
§Errors
Returns Error::ConnectionsStillOpen if any connection is open, and
Error::EngineCallFailed if the engine could not stop every thread. In the
latter case chdb.h documents the engine as closed for the process
regardless — “whether or not it manages to stop every thread” — so this
error means the engine is already shut down even though it is reported as
a failure; the crate’s internal shut-down flag is not set for it (see
SHUT_DOWN), so a subsequent open() will surface ConnectionFailed
rather than Error::EngineShutDown.
§Examples
let conn = chdb_rust::connection::Connection::open_in_memory()?;
drop(conn);
chdb_rust::runtime::shutdown()?;