Struct DbConnConfig 

pub struct DbConnConfig {

    pub engine: Option<DbEngineCfg>,
    pub dsn: Option<SecretString>,
    pub host: Option<String>,
    pub port: Option<u16>,
    pub user: Option<String>,
    pub password: Option<SecretString>,
    pub dbname: Option<String>,
    pub params: Option<HashMap<String, String>>,
    pub file: Option<String>,
    pub path: Option<PathBuf>,
    pub pool: Option<PoolCfg>,
    pub lock_keepalive: Option<Duration>,
    pub server: Option<String>,
}

Reusable DB connection config for both global servers and gears. DSN must be a FULL, valid DSN if provided (dsn crate compliant).

Not #[non_exhaustive]: callers (including first-party tests) build it with struct-literal + ..Default::default(), so that attribute would break them. Deserialization stays backward-compatible via #[serde(default)] on new fields, but adding a public field is a source-breaking change for exhaustive struct-literal callers — see the crate semver note.

Fields

engine: Option<DbEngineCfg>

Explicit database engine for this connection.

This is required for configurations without dsn, where the engine cannot be inferred reliably (e.g. distinguishing MySQL vs PostgreSQL, or selecting SQLite for file/path configs).

If both engine and dsn are provided, they must not conflict (validated at runtime).

dsn: Option<SecretString>

host: Option<String>

port: Option<u16>

user: Option<String>

password: Option<SecretString>

dbname: Option<String>

params: Option<HashMap<String, String>>

Backend-specific connection parameters.

Recognized keys are mapped to typed sqlx options (sslmode, application_name, statement_cache_capacity, …). Anything else is passed to PostgreSQL as a server runtime parameter, i.e. applied as if by SET on every connection in the pool.

That catch-all is how session GUCs are configured, and it is easy to miss because nothing names them:

params:
  statement_timeout: "5s"   # abort a single statement that runs longer
  lock_timeout: "2s"        # give up waiting for a row lock

Neither has a default here, so a deployment that sets nothing has neither bound. Note that statement_timeout bounds one statement, not a transaction. For the transaction as a whole there is transaction_timeout, added in PostgreSQL 17; before that version there is no equivalent, and idle_in_transaction_session_timeout bounds only the idle part. All three are runtime parameters, so all three are set the same way.

MySQL has no runtime-parameter equivalent; unrecognized keys there are rejected rather than forwarded.

file: Option<String>

path: Option<PathBuf>

pool: Option<PoolCfg>

lock_keepalive: Option<Duration>

Keepalive ping interval for the dedicated advisory-lock session (PG/MySQL only).

Defaults to crate::advisory_locks::DEFAULT_LOCK_KEEPALIVE when unset. Tune below any proxy/database idle-connection cutoff so the single lock session is not reaped.

server: Option<String>

Trait Implementations

impl Clone for DbConnConfig

fn clone(&self) -> DbConnConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for DbConnConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for DbConnConfig

fn default() -> DbConnConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for DbConnConfig

fn deserialize<__D>( __deserializer: __D, ) -> Result<DbConnConfig, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for DbConnConfig

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.