Database Management Operations
dbCreate/dbDestroy, dbLoad/dbUnload, dbOpen/dbClose—are performed asynchronously by default. This is generally advisable, as such operations are not atomic and may take a long time. In a large grid, loading a database may succeed immediately on many hosts, take a little longer on others, and much longer on others. Some hosts may, in fact, be down when the operation was run, so cannot perform the operation until they are back up.
By default, commands for these database operations return without waiting for completion, but they can optionally wait, with or without a timeout. With wait and a timeout, a command does not return until it has completed on all instances or reaches the timeout. With wait and no timeout, a command will never return if any instances are down. There are advantages and disadvantages to each approach, depending on factors such as how large the grid is. For a large grid, you may choose to proceed before the operation has completed on all instances, while on a small grid it may be more sensible to wait until it has completed on every instance.
TimesTen Scaleout tracks the state of a database, including each element of the database, and it is up to the user to check status of an operation (through the dbStatus command, optionally using the -all option for further details) to see how many instances have completed the operation. In particular, after loading the database, you can use the status information to determine if it has been loaded on enough instances for the database to be opened and users to start accessing it.
No command is provided to cancel an operation on any or all instances.