Controls the maximum time a single request can run. This option can be used to prevent a connection from consuming a significant amount of server resources for a long period of time.
Integer, 0 through 86400 (one day), in seconds
When this option is set to 0, requests do not time out.
Any request that takes longer than approximately request_timeout seconds (wall-clock time, not CPU time) is interrupted and an error is returned to the user. The error returned is SQLE_REQUEST_TIMEOUT: "Request interrupted due to timeout". If a request is blocked, and the blocking_timeout option is set to 0, then the request can remain blocked for a maximum of request_timeout seconds before returning a blocking error (for example, SQLE_LOCKED: "User '%1' has the row in '%2' locked").
User and public values 1 to 14 are not allowed. This prevents users from being locked out of the server if connecting takes a long time (for example, because of a complex login procedure).
This option can be used with both database client and HTTP/HTTPS requests. Note that setting the option in a stored procedure or HTTP/HTTPS request has no effect on the current request since the option value at the beginning of the request is used.
Setting the request_timeout public option should be done with caution as this can cause applications that have long running requests (such as dbvalid, dbbackup, and dbunload) to fail. Also, applications that do not use significant server resources, but that can block on another user can fail when request_timeout is set. One way to address these types of problems is to set the request_timeout option only for certain applications in the login procedure based on a connection's APPINFO value.
Setting this option may not prevent applications from using significant server resources if each request evaluates quickly, for example when fetching a result set containing many rows.